docs: mark daemon/client plan as implemented

pumpChild's PTY.Read raced Session.Shutdown's PTY.Close on the master
field (Close set it nil while Read read it). Benign at process exit on
main, but the daemon now runs Shutdown routinely (daemon stop). Guard
the field with a mutex, capturing the fd under the lock and doing the
blocking I/O outside it so Close still unblocks an in-flight Read.

Caught under: go test -race -run 'Daemon|NetClient|Owner' -count=5.

.gitea/workflows/release.yml

@@ -11,14 +11,19 @@ jobs:
     steps:
       - uses: actions/checkout@v4
 
-      - uses: actions/setup-go@v5
+      - uses: jdx/mise-action@v2
         with:
-          go-version-file: go.mod
           cache: true
 
-      - uses: mlugg/setup-zig@v1
+      - name: Cache Go modules
+        uses: actions/cache@v4
         with:
-          version: 0.15.2
+          path: |
+            ~/.cache/go-build
+            ~/go/pkg/mod
+          key: ${{ runner.os }}-go-${{ hashFiles('**/go.sum') }}
+          restore-keys: |
+            ${{ runner.os }}-go-
 
       - name: Build libghostty-vt
         run: make deps
@@ -30,7 +35,8 @@ jobs:
           CGO_ENABLED: 1
         run: |
           mkdir -p dist
-          go build -trimpath -ldflags="-s -w" \
+          go build -trimpath \
+            -ldflags="-s -w -X main.version=${{ github.ref_name }}" \
             -o dist/patterm-${{ github.ref_name }}-linux-amd64 \
             ./cmd/patterm
 

.gitignore

@@ -7,4 +7,5 @@ spike-report-*.txt
 /bin/
 /spike
 /.worktrees/
+/.claude/worktrees/
 internal/harness/.artifacts/

.mise.toml

@@ -0,0 +1,10 @@
+# mise config — `mise install` provisions the tools `make deps` needs.
+#
+# libghostty-vt is built from a pinned upstream Ghostty commit; that
+# commit's build.zig.zon pins minimum_zig_version = 0.15.2. We match
+# it here so contributors don't have to puzzle out the version from
+# a deep upstream file. The go pin matches go.mod so CI and local
+# builds use the same toolchain.
+[tools]
+zig = "0.15.2"
+go = "1.26.3"

CHANGELOG.md

@@ -6,6 +6,599 @@ loosely follows [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
 ## [Unreleased]
 
+### Added
+- `patterm daemon`, `patterm daemon stop`, and `patterm ls` now expose
+  a local unix-socket daemon lifecycle for the daemon/client split.
+- The local daemon protocol now supports attach, explicit detach,
+  project listing, focused-pane snapshots, pane chunks, resize/focus
+  updates, and daemon-owned command spawn requests while keeping child
+  processes alive after a client disconnects.
+- The default `patterm [dir]` startup now auto-starts the local daemon
+  on demand and attaches a thin terminal client over the unix-socket
+  transport; `--in-process` or `PATTERM_NO_DAEMON=1` keeps the legacy
+  single-process path available as an escape hatch.
+- `patterm daemon --listen HOST:PORT` can now opt into a TCP listener
+  for remote human clients, with the unix socket still enabled for
+  local clients.
+- `patterm connect --host HOST:PORT [--token TOKEN]` attaches the thin
+  client to a remote daemon over the same transport protocol.
+- TCP attaches now require a lightweight bearer token stored under
+  `$XDG_DATA_HOME/patterm/clients/token`; local unix-socket attaches
+  remain exempt and rely on socket file permissions.
+- The daemon now tracks a display owner per pane so a second client
+  viewing the same pane does not resize the underlying PTY/emulator;
+  ownership is released on detach and the next focuser can claim and
+  resize the pane.
+- patterm can now keep multiple local projects loaded in one loopback
+  daemon core, with command-palette entries to switch the current
+  client view or open another project without tearing down processes
+  in the previous project.
+- The status line now shows the current project name when multiple
+  projects are loaded, and the MCP startup greeting includes
+  `project_key` for diagnostics and future daemon routing.
+- MCP clients can now call `scratchpad_delete` with a scratchpad name
+  to remove a shared project scratchpad.
+
+### Changed
+- The tab bar now shows each visible agent tab's own summary instead
+  of only rendering the focused tab's summary.
+- Grid-mode `get_process_output` now returns whitespace-normalized
+  text to avoid sending padded terminal rows and repeated blank lines
+  over MCP.
+
+### Fixed
+- MCP scratchpad tools now route through the caller's project instead
+  of always using the daemon registry's default project.
+- Injected agent input now sends the submit Enter as a separated,
+  settled keystroke so messages reliably submit instead of sometimes
+  sitting unsent in the composer.
+- Codex agents are no longer reported idle while a turn is still
+  running.
+- Slow MCP tool calls such as `wait_for_pattern` no longer block later
+  tool calls on the same MCP connection.
+- Closing an agent now escalates from SIGTERM to SIGKILL when needed,
+  so agents that ignore SIGTERM disappear from the running tab bar
+  after one Close action while keeping their exited pane readable.
+- Sidebar timer indicators now repaint as their visible countdown
+  value changes, so labels progress from minutes to seconds without
+  waiting for unrelated terminal output or focus changes.
+- Raw terminal focused actions now show a single `Close` row instead
+  of separate stop/delete-style lifecycle choices that did the same
+  thing for ephemeral terminal panes.
+- Restarting a process from the palette now restores the focused pane
+  and host chrome before waiting for the old process to exit, so the
+  tab bar and sidebar do not disappear during slow restarts.
+- Deleting the focused scratchpad now moves focus to another
+  scratchpad when one exists, or back to a running terminal/agent
+  instead of dropping into the empty state.
+- Multiline paste into raw terminal and command panes no longer pays
+  the agent-specific per-Enter delay, making large pasted input arrive
+  as one PTY write outside Claude/Codex/OpenCode panes.
+
+## [0.0.7] - 2026-05-18
+
+### Added
+- The top tab bar now prefixes each agent tab's label with its
+  idle-state glyph (✕ error, ? permission, ◐ thinking, ○ idle, ●
+  working), matching the sidebar's vocabulary so the state of every
+  open agent is visible without opening or focusing each tab.
+
+### Changed
+- Built-in agent presets (`claude`, `codex`, `opencode`) now live in
+  memory and user preset files merge over them by name instead of
+  patterm writing default preset files into `$XDG_CONFIG_HOME`. Add
+  `"disabled": true` in a matching user preset to hide a built-in.
+- Generated MCP config files for agent launches now live under the
+  runtime agent directory instead of `$XDG_CONFIG_HOME/patterm/mcp`.
+- Auto-summarization settings now save as soon as a changed row is
+  applied, including cadence/provider/toggle changes and model edits,
+  without requiring a separate save step.
+- The Agents / Auto-summarization settings screen no longer shows
+  explicit Save, Cancel, or Back rows, and its footer copy no longer
+  describes a separate save/cancel flow.
+- Auto-summarization setting rows now visually separate grey labels
+  from regular-colour values.
+- The active-thread summary in the tab bar is now constrained to the
+  active tab's width instead of spanning the whole top row.
+- Sidebar summary text now wraps from the full summary text instead of
+  using an ellipsized single-line value.
+
+### Fixed
+- Claude permission prompts are now detected from the rendered pane as
+  well as the recent output tail, so the sidebar marks the pane as
+  waiting for permission even while `Calling patterm...` continues to
+  repaint.
+- Removed the redundant "Back to Settings" row from the
+  Agents / Auto-summarization settings screen.
+- Pending `timer_*` entries are now cancelled when their owning or
+  watched child is closed via `close_process`, preventing stale
+  timer bodies from being re-delivered to the orchestrator pane
+  after the work has already been handled.
+
+## [0.0.6] - 2026-05-15
+
+### Changed
+- Toast notifications now reserve three content rows and word-wrap
+  the message body inside the box, replacing the previous
+  single-line+ellipsis layout. The `Ctrl-N · N more` inline hint is
+  gone; instead the host status strip surfaces a `Ctrl-N · dismiss`
+  hint, shown only while a notification is on screen so the chord
+  doesn't advertise itself when it has nothing to dismiss.
+
+### Fixed
+- Auto-summary no longer fails immediately with `codex summarizer:
+  error: unexpected argument '--ask-for-approval' found`. The codex
+  CLI dropped that flag; we now rely on `--sandbox read-only` (which
+  already implies no approvals) instead of passing it.
+- Toast box no longer flickers / half-erases while the focused
+  child (claude, codex, opencode, etc.) repaints its TUI. The
+  overlay is now stitched onto the end of the per-chunk PTY write
+  under `outMu`, and wrapped in DECSET 2026 (synchronized output)
+  brackets so terminals that support it batch the child's redraw +
+  the box paint into a single frame instead of racing cell-by-cell.
+
+## [0.0.5] - 2026-05-15
+
+### Changed
+- Replaced the single-slot status-line "flash" with a stackable toast
+  surface anchored at the top-right of the focused pane. `flashError`,
+  `flashTransient`, and MCP `request_human_attention` now push onto
+  the toast stack (cap 5, oldest drops). Toasts persist until
+  dismissed with `Ctrl-N`, or cleared via the new
+  "Clear notifications" palette command. The status line no longer
+  shows the `[!]` prefix.
+- `Ctrl-N` is consumed by the host only when there is a toast to
+  dismiss; an empty stack lets `Ctrl-N` pass through to the focused
+  child so readline / nano / emacs / opencode keep their bindings.
+- Command palette is calmer when something is focused. Focused-section
+  rows now read as bare verbs (`Rename`, `Close`, `Stop`, `Restart`,
+  `Delete`, `Edit`) instead of repeating the focused name (`Close
+  agent: codex`); the title bar's `on: codex` / `pad: notes.md`
+  carries the subject. Fuzzy queries still match the dropped context
+  through the row hint (e.g. typing `close codex` still finds the
+  Close row).
+- Dashed `── Focused ──` / `── Open ──` / `── Spawn ──` section
+  banners are gone. Sections are separated by a single blank spacer
+  row, so the action labels themselves carry the visual weight.
+- The Open section no longer lists a `Switch to <current>` row for
+  the pane you're already focused on.
+
+## [0.0.4] - 2026-05-15
+
+### Changed
+- Release workflow (`.gitea/workflows/release.yml`) now provisions
+  Zig and Go through `jdx/mise-action@v2`, reading the versions from
+  `.mise.toml` (zig 0.15.2, go 1.26.3). Both toolchains were
+  previously installed via `mlugg/setup-zig` and `actions/setup-go`,
+  whose mirror chase / GitHub fetch combined for ~8 minutes per run
+  before any patterm code compiled. mise pulls each tool once and
+  caches the install dir, so subsequent runs hit the cache instead of
+  re-downloading. `make deps` still resolves zig via `mise which zig`
+  with a PATH fallback; `go.mod` already pinned `go 1.26.3`, so the
+  new `go` entry in `.mise.toml` just keeps CI and local builds on
+  the same toolchain.
+- A Go module/build cache step (`actions/cache@v4`, keyed on
+  `go.sum`) was added so `go build` doesn't re-download dependencies
+  on every tag push.
+
+## [0.0.3] - 2026-05-15
+
+### Added
+- Auto-summarization for top-level agent tabs. patterm now loads
+  `$XDG_CONFIG_HOME/patterm/settings.json`, enables Codex-based
+  summaries by default (`gpt-5.4-mini`; OpenCode defaults to
+  `opencode-go/minimax-m2.7`), and can run Codex, OpenCode, or opt-in
+  Claude summarizers with configurable model names. Summary
+  attempts are armed by meaningful human input, wait for recent output
+  to go quiet, and respect a minimum cadence so unchanged tabs are not
+  summarized on a timer. The active thread summary appears under the
+  top tab title and in the sidebar below the Agent Tree section.
+- Settings overlay reachable from the command palette via
+  `Open Settings`. The searchable Settings picker opens
+  `Agents / Auto-summarization`, where users can enable/disable
+  summaries, choose provider, edit provider model names, cycle cadence,
+  test the selected summarizer (`patterm okay`), summarize the current
+  top-level agent immediately, and explicitly save or cancel draft
+  settings changes. Cadence choices match Solo: `15s`, `30s`, and
+  `1m`; the value is a minimum quiet/activity gap before another
+  summary attempt for the same top-level agent, not a background
+  periodic timer.
+
+### Changed
+- Command palette UX overhaul. The single flat list grew section
+  bands (`── Focused ──`, `── Open ──`, `── Spawn ──`, `── Quit ──`)
+  so the rows are scannable at a glance; cursor navigation skips
+  the dim header rows transparently. A chip strip — `[All] Open
+  Spawn Close` — sits below the query line and tracks the active
+  macro filter; `Tab` / `Shift-Tab` cycle through the chips, and
+  the typed-prefix macros (`sw `, `sp `, `k `) still work and now
+  collapse the whole prefix on a single backspace instead of
+  leaving a stray `sw` behind. The title bar surfaces the current
+  focus subject (`on: <child>` / `pad: <name>`) so the user knows
+  which Focused row is targeting what. The duplicate global Close
+  list is gone — close is reachable via the Focused-section action,
+  the `k ` macro / `[Close]` chip, or the new `Ctrl-X` inline close
+  on a Switch row. The "(current)" marker on the focused Switch row
+  became a leading `▶`. The empty-state hint now reads `no matches
+  · ⌫ to widen` instead of bare `no matches`. The middle divider
+  shows a `▼ N more` / `▲ N above` scroll indicator when the list
+  overflows, and the footer carries a `cursor/total` counter.
+- Spawn verbs are unified on **Spawn**: `Run process: …` →
+  `Spawn process: …`, `New Terminal` → `Spawn terminal`, and the
+  freeform-form row is now `Spawn process… (custom)` so the
+  trailing ellipsis still signals it opens a form.
+- Filtering switched from binary fuzzy-include to scored ranking.
+  Prefix matches beat word-boundary matches beat substring matches
+  beat scattered-fuzzy matches; ties fall back to section order so
+  a Focused-section hit always outranks an equally tight Spawn
+  hit. The matched characters in the rendered label render in
+  accent+bold so the user can see why a row matched.
+- Rename forms split the long subject (`scratchpad:
+  some-really-long-name.md`) onto its own dim row above the input
+  so the title bar no longer truncates with an ellipsis when the
+  subject name is wide.
+- New palette accelerators: `Alt-1` … `Alt-9` quick-pick the Nth
+  visible row, `Home` / `End` jump to first / last selectable row,
+  `?` (with empty query) opens an inline keybinding cheat-sheet
+  which any further keystroke dismisses, and `Ctrl-R` inside the
+  Spawn-process form toggles "Relaunch on exit" without leaving
+  the command field.
+
+### Fixed
+- Error/status flashes now restore the currently focused pane instead
+  of drawing the empty-state hint over a running agent or process.
+- Release workflow (`.gitea/workflows/release.yml`) now uses
+  `mlugg/setup-zig@v2` instead of the deprecated `@v1`. v1 hard-coded
+  the pre-0.14 tarball name (`zig-linux-x86_64-<ver>.tar.xz`), so
+  every mirror and the official `ziglang.org/builds` returned 404 for
+  Zig 0.15.2 and the v0.0.1 / v0.0.2 tag pushes never produced a
+  release asset. v2 uses the post-0.14 `zig-x86_64-linux-<ver>.tar.xz`
+  layout, so the runner can fetch Zig and build patterm.
+- Typing into a focused child while its emulator viewport is
+  scrolled up into scrollback history now auto-snaps the viewport
+  back to the live area. Previously the keystroke reached the
+  child PTY but the input box was off-screen below the visible
+  region, so it looked like typing did nothing. Wheel scrolling
+  and Ctrl-B are unchanged; only forwarded keystrokes snap.
+- Top tab bar now keeps the top-level agent's tab highlighted
+  when focus is on one of its sub-agents (or on a Processes pane
+  entry, matching the existing agent-tree behavior). Previously
+  the tab would lose its highlight as soon as you stepped into a
+  child agent, even though you were still within that thread.
+
+### Changed
+- MCP tool descriptions and `help('coordination')` /
+  `help('readiness')` now spell out that a sub-agent's reply to
+  `send_message` lands in the caller's own pane (tagged
+  `[sub-agent:<name>]`), not in the sub-agent's output. The canonical
+  wait-for-reply pattern — `send_message` → `timer_fire_when_idle_any`
+  on the sub-agent → read your own pane — is now called out on
+  `send_message`, `wait_for_pattern`, both `timer_fire_when_idle_*`,
+  the help topics, and the server-instructions preamble every agent
+  reads at startup. Previously `wait_for_pattern` was the obvious
+  blocking primitive in the catalog, and agents routinely called it
+  against the sub-agent for a reply that had already arrived in their
+  own pane, deadlocking until the wait timed out. No behaviour
+  changes; descriptions only.
+- Agent-initiated `spawn_agent` and `spawn_process` MCP calls no
+  longer steal viewport focus from the currently active tab. The
+  new child still appears in the sidebar and tab bar; switch to it
+  explicitly via the palette or `select_process`. Palette-initiated
+  spawns and persistence restores are unchanged — they still auto-
+  focus the new pane.
+- Sidebar rows (Processes, Agent Tree, Scratchpads) now truncate
+  overflowing names with a trailing `…` instead of spilling into
+  the main viewport. The focused row marquees its name when it
+  overflows — 1 s hold on the head, ~150 ms per cell scroll until
+  the tail is visible, 1 s hold on the tail, snap back. Row
+  position never moves while the marquee animates. When budget is
+  tight, the trailing timer indicator drops before the name
+  ellipses, since the name is the only identifier the row carries.
+
+## [0.0.2] - 2026-05-15
+
+### Added
+- `.mise.toml` pinning `zig = "0.15.2"` (the minimum version the
+  vendored Ghostty commit requires). Contributors run
+  `mise install` once; the Makefile picks up the resulting `zig`
+  binary automatically via `mise which zig` and falls back to
+  PATH when mise isn't available, so the existing build flow
+  still works.
+- ASCII-video stress benchmarks (`internal/app/bench_test.go`):
+  per-frame and per-stream variants at 30 / 60 / 120 fps targets,
+  three workload fixtures (8-colour cells, 24-bit truecolor cells,
+  and a Bad-Apple-style 1-bit pattern). Each stream benchmark
+  reports `µs/frame`, an achievable `fps_ceiling`, and `budget_pct`
+  so you can read off "do we hit N fps?" directly. A matching
+  Pipeline_ASCIIVideo_* set includes libghostty-vt's em.Write CGO
+  and an io.Discard stdout write so the FPS claim reflects the
+  whole pipeline, not just the renderer.
+- MCP `initialize.instructions`, the `spawn_agent` tool description
+  (visible to LLMs via `tools/list`), and the `help('spawning')`
+  topic now spell out — in the three places vendor TUIs actually
+  consult — that the connected `patterm` MCP server is the only
+  correct way to drive the host. Anti-patterns called out by name:
+  (a) trying to launch `patterm` / `patterm mcp-stdio` themselves,
+  (b) piping JSON-RPC into the per-PID Unix socket via `perl` /
+  `nc` / `socat` / `curl`, and (c) shelling out to `claude` /
+  `codex` / `opencode` to start a peer. Each of those bypasses
+  caller identity, so a sub-agent spawned that way reads back as
+  a stray top-level tab instead of a child under the spawning
+  agent. Codex was hitting (b) and (c) in practice — this is the
+  fix.
+- `--debug[=DIR]` flag captures detailed run artefacts for offline
+  analysis: a verbose `patterm.log` (the existing `PATTERM_DEBUG_LOG`
+  stream), an `events.jsonl` lifecycle log (spawn / exit / idle-state
+  changes with timestamps), and per-child `<id>.raw` files containing
+  the raw PTY byte stream. With no argument, the dated subdir
+  `$XDG_STATE_HOME/patterm/debug/YYYYMMDD-HHMMSS` is used; pass an
+  explicit path to override. All output goes to files — stdout/stderr
+  are untouched.
+- `--profile[=DIR]` flag captures pprof data plus concrete
+  performance counters for performance work: `cpu.pprof` (running
+  for the lifetime of the session), plus `heap.pprof` and
+  `goroutine.pprof` snapshots written on shutdown; alongside them,
+  a per-hot-path metrics tracker writes `metrics.jsonl` (one row
+  per second with chunk/byte rates, per-stage mean and max
+  latencies, and cache hit rates) plus a final `metrics.json`
+  aggregate and a human-readable `summary.txt` on exit.
+  Instrumented hot paths: `OnPTYOut`, viewport `renderer.Render`,
+  host stdout writes, libghostty-vt `emulator.Write` / `Title`,
+  sidebar / tab bar / status line draws (with cache-hit
+  accounting), snapshot replays, and the chrome ticker (so you can
+  see how often it fires with nothing to do). Defaults to
+  `$XDG_STATE_HOME/patterm/profile/YYYYMMDD-HHMMSS`. All
+  diagnostics (startup, errors) are written to `profile.log`
+  inside the dir, never to stdout/stderr.
+- Renderer benchmark suite (`internal/app/bench_test.go`). Three
+  workload fixtures — plain ASCII, SGR-styled lines, and a
+  ratatui-style cursor-shuffling burst — plus an OSC-gate
+  micro-benchmark. Run via `go test -bench=. -benchmem
+  ./internal/app/`. Gives a stable reference for the per-chunk
+  cost of the viewport renderer so future changes can be compared
+  apples-to-apples.
+- "New Terminal" entry in the command palette spawns a bare interactive
+  `$SHELL` pane (kind `terminal`). Unlike "Run process: …" presets,
+  which are session-persistent and reachable via `restart_process`,
+  terminals are ephemeral — once they exit they vanish from the
+  Processes sidebar instead of lingering as a dead row. The default
+  `shell` process preset that previously seeded on first run has been
+  removed; this entry replaces it.
+- Per-child idle-state classifier with five states (`idle`, `working`,
+  `thinking`, `permission`, `error`) and three pluggable strategies:
+  `output_activity` (claude / opencode defaults), `osc_title_stability`
+  (codex), and `osc_title_status` (gemini-style status-in-title agents).
+  Optional `permission_patterns` / `thinking_patterns` / `error_patterns`
+  regexes promote a base state when matched against the tail of recent
+  output. State and last-match reason are exposed via MCP on
+  `ProcessInfo` and `get_process_status` (`idle_state`, `idle_reason`).
+- New `idle_detection` block on `preset.Preset` for setting the strategy
+  threshold, title-to-state map, and promoter regex lists. Bundled
+  defaults are shipped for the first-party claude / codex / opencode
+  presets.
+- Sidebar now renders a state glyph per process row (`○` idle, `●`
+  working, `◐` thinking, `?` permission, `✕` error) and, when a process
+  has a pending or paused timer, appends a nearest-timer indicator
+  (`⏱ 12s` or `⏸ paused`).
+- MCP timer surface expanded to match Solo's tool set: `timer_set`,
+  `timer_fire_when_idle_any`, `timer_fire_when_idle_all`, `timer_cancel`,
+  `timer_pause`, `timer_resume`, `timer_list`. Idle-aware timers
+  registered against already-idle children fire synchronously
+  (`status: already_satisfied`) for `idle_all`, and report
+  `already_idle` / `waiting_on` arrays so callers can introspect the
+  watch set. Timer bodies are delivered to the owner process via the
+  same orchestrator-injection path as `send_message`.
+- Timer tools accept an explicit `owner_process_id` so top-level
+  (non-agent) callers — including the harness MCP client — can attribute
+  timers to a specific process. Omitting it treats the caller as the
+  orchestrator with universal cancel / pause / resume / list privileges.
+- libghostty-vt `Title()` accessor on the emulator surface, polled from
+  the session pump so OSC 0/1/2 title updates feed into the classifier
+  without a callback round-trip.
+- Harness `wait_until_mcp` step type that re-runs an MCP method until an
+  assertion (Equals / Contains) holds or the timeout elapses. Used by
+  the new idle / timer scenarios.
+- User-created top-level command processes now survive a patterm
+  restart. Each spawn (palette form, command preset, or MCP
+  `spawn_process` with `kind=command`) writes a record to
+  `$XDG_DATA_HOME/patterm/projects/<key>/processes.json`; on next
+  startup patterm replays those entries before the UI accepts input,
+  so things like `bun run dev` or `tail -F log` come back without
+  re-typing. `close_process` (and the palette's close action) drops
+  the entry, and rename / "relaunch on exit" toggles are mirrored as
+  they happen. Agents and terminals stay ephemeral by design.
+- The command palette (Ctrl-K) now surfaces context-aware actions at
+  the top of the list, based on what's currently focused:
+  - Scratchpad in focus: `Delete`, `Rename` (inline form), and `Edit`
+    (fire-and-forget launch of `zed` against the pad file).
+  - Agent in focus: `Rename agent` (inline form) and `Close agent`.
+  - Process in focus: `Rename process`, `Delete process` (drops the
+    entry; SIGKILLs if alive), `Stop process` (SIGTERM, keep entry),
+    and `Restart process` (same argv).
+- `patterm --version` prints the build version, git commit, and build
+  date (e.g. `patterm v0.0.1 (commit abc1234, built 2026-05-14)`). The
+  version string is injected by the build (`make patterm` derives it
+  from `git describe`; the release workflow injects the pushed tag).
+  Commit and date come from the Go toolchain's embedded VCS info, so
+  nothing has to be bumped by hand.
+- Ctrl+R restarts the focused command process from the Processes
+  sidebar, including command entries that have already exited.
+- Scratchpads are now first-class navigation targets. Ctrl+W / Ctrl+S
+  step from the Processes section and agent tree onto scratchpad
+  entries; a focused scratchpad renders its content in the main
+  viewport (with the pad name as the header) instead of cramping the
+  bottom of the sidebar. The sidebar's scratchpad section is a names-
+  only list with the focused pad highlighted, and external MCP
+  `scratchpad_write` / `scratchpad_append` updates repaint the pad
+  view immediately.
+- Focused scratchpads now render as markdown — headings, bold, inline
+  code, fenced code blocks, bullet/numbered lists, blockquotes, and
+  horizontal rules pick up styling instead of the previous flat
+  word-wrap. Long pads scroll: the mouse wheel is the primary control
+  (patterm enables SGR mouse reporting while a pad is focused), and
+  Up/Down / PageUp/PageDown / Home / End work for keyboard users. The
+  header reports the visible row range and total row count. Esc leaves
+  the pad view and falls back to the first running process (or an
+  empty viewport). The scroll offset is preserved across MCP
+  `scratchpad_write` / `scratchpad_append` writes so a live update
+  doesn't snap the view back to the top.
+- Inline wheel scrollback for the focused child, backed by
+  libghostty-vt's native 5000-row scrollback history. On the primary
+  screen, mouse-wheel events scroll the emulator viewport in-place with
+  full SGR styling preserved — no modal view to enter or exit. On the
+  alternate screen wheel events still pass through to the child so
+  vim / less / codex receive them as input. Ctrl+B snaps the viewport
+  back to the live (bottom) area as the escape hatch from a scrolled-up
+  state. Patterm now keeps SGR mouse reporting armed on the host
+  terminal while the alt screen is active and filters mouse-mode
+  toggles from the child stream so wheel events keep arriving even
+  after a child program disables mouse tracking.
+
+### Changed
+- The palette's per-child "Kill <name>" action is now labelled
+  "Close <name>". The underlying signal (SIGTERM) and behaviour are
+  unchanged; the new label matches the existing "Close agent: …"
+  context entry and reads less violent for what is really just a
+  graceful termination.
+- `timer_wait` is now a thin wrapper over the shared timer manager
+  (`timer_set` semantics). Existing callers see no behavioural change;
+  the timer is visible in `timer_list` while it's pending.
+- CLI flag parsing switched from Go's stdlib `flag` to `spf13/pflag`.
+  `--project` (and the internal `--socket` / `--identity` /
+  `--scenario` / `--patterm-bin` flags) are now the only accepted form
+  — single-hyphen long flags like `-project` are rejected. Help output
+  renders the canonical `--flag` form.
+
+### Fixed
+- `make deps` now builds libghostty-vt with `-Doptimize=ReleaseFast`
+  instead of zig's silent `Debug` default, and resolves `zig`
+  through `mise` when a project `.mise.toml` pins it. The
+  default-Debug build shipped an unoptimised CSI/SGR parser that
+  ate 16-29 ms per 30-70 KiB full-screen frame in benchmarks,
+  capping the entire PTY-to-host pipeline at 34-63 fps. After the
+  rebuild the same pipeline runs at **930-2030 fps**: 27-32× the
+  prior throughput, and 7-16× margin over 120 fps for full-screen
+  truecolor ASCII video. Static library size drops from 33 MiB to
+  13 MiB. Override with `make deps GHOSTTY_VT_OPTIMIZE=Debug` only
+  when debugging the upstream library itself. Apply on existing
+  checkouts with `mise install && make clean-deps && make deps`.
+- Long claude session resume (and codex steady-state rendering) is
+  noticeably faster. Two costs that scaled per-PTY-chunk are now
+  deferred or short-circuited: (1) `drawSidebar()` used to run
+  synchronously for every chunk that scrolled — on a session
+  resume where every chunk scrolls, this rebuilt the full sidebar
+  string hundreds of times for a frame that was almost always
+  cache-equal. The sidebar now signals dirty and the chrome ticker
+  (60 Hz) handles the repaint. (2) `pumpChild` polled the
+  emulator's OSC title after every PTY chunk via CGO, even for
+  chunks (the common case under codex/ratatui) that carry no OSC
+  bytes at all. The poll is now gated on a containsOSC scan over
+  the chunk.
+- Click-and-drag text selection from alt-screen TUIs (codex in
+  particular) now works. Patterm used to keep host SGR mouse
+  reporting armed continuously, which forced the host terminal to
+  forward every click as an escape sequence and prevented native
+  selection. The host's mouse mode now follows the focused child's
+  screen side: primary-screen children keep mouse armed (so wheel
+  scrollback works), alt-screen children get host mouse disabled by
+  default. Alt-screen TUIs that need mouse events (vim, less, etc.)
+  re-enable mouse-mode themselves; the viewport renderer forwards
+  those toggles to the host while the child is on alt. Leaving alt
+  re-arms host mouse reporting so wheel scrollback resumes.
+- Exited terminal panes (kind `terminal`, including those launched via
+  the new "New Terminal" palette entry or MCP `spawn_process` with
+  `kind=terminal`) are now removed from the session and the Processes
+  sidebar as soon as they exit. Previously they stuck around as a
+  greyed-out row indistinguishable from an exited command process,
+  even though terminals have no restart path.
+- `whoami` and `help("timers")` now advertise the full Solo-parity timer
+  surface (`timer_set`, `timer_fire_when_idle_any`,
+  `timer_fire_when_idle_all`, `timer_cancel`, `timer_pause`,
+  `timer_resume`, `timer_list`) so agents using either tool for
+  orientation discover them — previously only `timer_wait` was listed.
+- Resuming a paused idle-aware timer now re-checks the satisfaction
+  condition. Previously, if every watched process became idle (or, for
+  `idle_any`, any non-baseline watcher went idle) while the timer was
+  paused, the timer stayed pending forever because no further state
+  transitions were observed.
+- Fired and canceled timers are now removed from the timer registry,
+  so long-running patterm sessions no longer accumulate completed
+  timer records and message bodies. `timer_list` and the sidebar
+  indicator already filtered them out; only the in-memory leak is
+  fixed.
+- Per-preset idle-detection config is now installed through `SpawnSpec`
+  before the child is published to the session, closing a race in
+  which the classifier goroutine could observe a freshly spawned
+  process before its preset's classifier strategy was attached.
+- Opening the command palette while a scratchpad was focused left the
+  palette wedged — typing did nothing and Esc left the palette's top
+  border drawn over the pad until you closed the pad with Ctrl-W and
+  re-opened the palette. The stdin loop's scratchpad-input branch ran
+  before the palette branch and silently dropped every byte except a
+  handful of app-level chords, so palette filter input and Esc never
+  reached `palette.handleInput`. The palette branch now takes
+  precedence whenever the palette is open, and `closePalette` repaints
+  the pad (instead of the empty focused-child slot) on cancel / no-op
+  action. Switching from a pad to a child via the palette now clears
+  the pad focus and wipes the viewport, matching `focusProcess`.
+- Tab bar and bottom status row no longer get overwritten by long
+  claude / codex sessions. Three holes were letting child output land
+  on the chrome: (1) absolute cursor moves — CUP / HVP / VPA — added
+  the row offset but didn't clamp to the viewport, so a child whose
+  internal row state drifted past its assigned height could walk the
+  host cursor onto the status row (or above the tab bar); (2) relative
+  cursor moves — CUU / CUD / CNL / CPL — were forwarded verbatim, so
+  a `CSI 50 A` from viewport row 1 walked the host cursor into the
+  tab bar before the next printable wiped it; (3) the host's DECSTBM
+  scroll region was only set during snapshot-replay preludes, so the
+  windows between (startup before first focus, post-SIGWINCH,
+  post-clearScreen) left the region defaulted to the full screen and
+  any LF / IND / NEL / RI / SU / SD at the viewport bottom scrolled
+  the chrome rows along with the pane. The cursor shifter now clamps
+  CUP/HVP/VPA rows to mainTop..mainBottom, the viewport renderer
+  rewrites CUU/CUD/CNL/CPL with a clamped step (and homes the column
+  for CNL/CPL), and patterm installs the host scroll region after
+  `enterScreen` and after every `clearScreen` (and resets it cleanly
+  on `leaveScreen` so the calling shell isn't left with a constrained
+  region).
+- Plain line-feed scrolling at the bottom of a child pane now invalidates
+  and repaints the sidebar, so long agent output can no longer drag the
+  sidebar border and labels out of view while the chrome cache stays warm.
+- Child DEC origin-mode and left/right-margin controls are now handled
+  inside the viewport renderer instead of being forwarded to the host
+  terminal, so later tab bar, sidebar, and status-line repaints keep
+  using physical screen coordinates and do not appear inside the
+  focused pane.
+- Exited command processes in the top Processes section are now reachable
+  with Ctrl+W/S navigation, so a dead shell entry can be focused and
+  restarted instead of becoming a visible but unreachable row.
+- Resizing the host terminal no longer makes codex (and other
+  diff-based TUIs) scroll-jump for several seconds. SIGWINCH is now
+  coalesced into a single resize after an ~80ms idle, the resize path
+  skips the full snapshot replay (the child's own SIGWINCH-driven
+  redraw fills the viewport), and `Child.NudgeRedraw` no longer
+  toggles the PTY through rows-1 + rows back-to-back during a
+  drag-resize.
+- Steady-state CPU during a long codex session dropped sharply.
+  Tab-bar and status-line repaints moved off the per-PTY-chunk path
+  to a 16ms chrome ticker; the scratchpad listing is cached and only
+  rebuilt when the pads change; the post-spawn / post-repaint
+  styled-snapshot replay budget dropped from 8 chunks to 2; URL/port
+  scanning short-circuits chunks that don't contain "http"; the
+  three writes around the autowrap toggle in `OnPTYOut` collapsed
+  into one syscall; the per-PTY-read `make+copy` was removed (the
+  64 KiB read buffer is reused, with a documented "do not retain"
+  listener contract); session listeners now dispatch through an
+  `atomic.Pointer` snapshot instead of a mutex copy on every chunk;
+  the per-child output ring is a true wrap-around buffer instead of
+  a slide-and-trim slice; `wait_for_pattern` wakes on PTY chunk
+  events with a 500ms fallback instead of unconditional 50ms
+  polling; ANSI stripping in MCP `get_process_output stream`,
+  `search_output`, and `wait_for_pattern scrollback` is now an
+  in-place byte walk; and the viewport renderer copies long ASCII
+  runs en bloc instead of feeding the state machine one byte at a
+  time.
+
 ## [0.0.1] - 2026-05-14
 
 ### Fixed

Makefile

@@ -20,10 +20,30 @@ $(SOURCE)/.git/HEAD:
 
 deps-fetch: $(SOURCE)/.git/HEAD
 
+# Zig's `standardOptimizeOption` defaults to .Debug when no
+# -Doptimize is passed, which makes libghostty-vt's CSI/SGR parser
+# an order of magnitude slower — truecolor full-screen frames spend
+# ~16-29 ms each in em.Write under Debug (see
+# internal/app/bench_test.go BenchmarkEmulator_Write_*), which caps
+# the full PTY-to-host pipeline at ~60 fps. ReleaseFast is the
+# right default for the shipped artefact. Override with
+# `make deps GHOSTTY_VT_OPTIMIZE=Debug` when you actually want a
+# debug build of the upstream lib.
+GHOSTTY_VT_OPTIMIZE ?= ReleaseFast
+
+# Resolve zig via the project's mise pin (.mise.toml) when available,
+# falling back to whatever's on PATH. mise keeps the zig version in
+# lockstep with what the pinned ghostty commit requires; without it,
+# contributors have to chase the version requirement themselves.
+ZIG := $(shell command -v mise >/dev/null && mise which zig 2>/dev/null || command -v zig 2>/dev/null)
+
 $(INSTALL)/lib/libghostty-vt.a: $(SOURCE)/.git/HEAD
-	@command -v zig >/dev/null || { echo "ERROR: zig not on PATH (need >=0.15.2 to build libghostty-vt)"; exit 1; }
-	@echo ">> building libghostty-vt with zig"
-	@cd $(SOURCE) && zig build -Demit-lib-vt --prefix $(INSTALL)
+	@if [ -z "$(ZIG)" ]; then \
+		echo "ERROR: zig not available. Run \`mise install\` (see .mise.toml — needs zig 0.15.2) or install zig manually."; \
+		exit 1; \
+	fi
+	@echo ">> building libghostty-vt with $(ZIG) (optimize=$(GHOSTTY_VT_OPTIMIZE))"
+	@cd $(SOURCE) && $(ZIG) build -Demit-lib-vt -Doptimize=$(GHOSTTY_VT_OPTIMIZE) --prefix $(INSTALL)
 	@test -f $(INSTALL)/lib/libghostty-vt.a || { echo "ERROR: expected static lib at $(INSTALL)/lib/libghostty-vt.a"; exit 1; }
 	@echo ">> libghostty-vt installed under $(INSTALL)"
 
@@ -32,11 +52,13 @@ deps-build: $(INSTALL)/lib/libghostty-vt.a
 clean-deps:
 	rm -rf $(SOURCE) $(INSTALL)
 
+VERSION ?= $(shell git describe --tags --always --dirty 2>/dev/null || echo dev)
+
 spike: deps
 	go build -o ./bin/spike ./cmd/spike
 
 patterm: deps
-	go build -o ./bin/patterm ./cmd/patterm
+	go build -ldflags "-X main.version=$(VERSION)" -o ./bin/patterm ./cmd/patterm
 
 test: deps
 	go test ./...

SPEC.md

@@ -39,7 +39,7 @@ The tool is one Go process that owns: the TUI, all PTYs, vt-emulated grids, sess
 
 ## 3. Project state layout
 
-Scratchpads (user data) live under `$XDG_DATA_HOME`; presets and config live under `$XDG_CONFIG_HOME`.
+Scratchpads (user data) live under `$XDG_DATA_HOME`; user-authored preset overlays and config live under `$XDG_CONFIG_HOME`.
 
 ```
 $XDG_DATA_HOME/patterm/
@@ -53,12 +53,12 @@ $XDG_DATA_HOME/patterm/
             └── <agent-written>.md
 
 $XDG_CONFIG_HOME/patterm/
-├── config.json                # global settings (theme, default keymap, etc.)
+├── settings.json              # global settings, written only after the user changes settings
 └── presets/
     ├── agents/
-    │   ├── claude.json        # ships as default
-    │   ├── codex.json         # ships as default
-    │   ├── opencode.json      # ships as default
+    │   ├── claude.json        # optional overlay for built-in claude
+    │   ├── codex.json         # optional overlay for built-in codex
+    │   ├── opencode.json      # optional overlay for built-in opencode
     │   └── <user-defined>.json
     └── processes/
         ├── dev.json           # e.g. { "name": "bun run dev", "argv": ["bun", "run", "dev"] }
@@ -66,7 +66,7 @@ $XDG_CONFIG_HOME/patterm/
         └── <user-defined>.json
 ```
 
-Both preset directories are scanned at startup; every file found becomes a palette entry ("Spawn agent: claude", "Run process: bun run dev", …). Presets are project-agnostic in v1 — the same set is available in every project. Per-project overrides can be added later.
+patterm always has built-in agent presets for `claude`, `codex`, and `opencode`. User preset files are scanned at startup and merged into matching built-ins by `name`, or added as standalone custom presets when the name is new. A matching file with `"disabled": true` hides a built-in. Startup does not write default preset files. Presets are project-agnostic in v1 — the same set is available in every project. Per-project overrides can be added later.
 
 Project key = `sha256(realpath(project_dir))[:16]`. Used only as a scratchpad directory name — there is no daemon to look up.
 
@@ -121,7 +121,7 @@ Scratchpads and command-preset trust grants persist across runs. Sessions and ch
 Almost all application functions are driven through a single command palette opened with `Ctrl-K`. The palette is a fuzzy-searchable list of commands, scoped to whatever makes sense for the current focus. Two kinds of entries appear:
 
 - **Built-in commands** — "Switch to session…", "Focus pane…", "Take input control", "Release control to orchestrator", "Open scratchpad…", "Kill child…", "Quit", etc.
-- **Preset commands** — one entry per file under `$XDG_CONFIG_HOME/patterm/presets/`. Agent presets surface as "Spawn agent: codex" / "Spawn agent: claude" / …; process presets surface as "Run process: bun run dev" / "Run process: vitest" / …. The label comes from the preset's `name` field; the action is "launch this preset into a new pane."
+- **Preset commands** — one entry per built-in or user-defined preset. Agent presets surface as "Spawn agent: codex" / "Spawn agent: claude" / …; process presets surface as "Run process: bun run dev" / "Run process: vitest" / …. The label comes from the preset's `name` field; the action is "launch this preset into a new pane."
 
 Selecting a preset either launches it immediately (no required args) or opens a sub-palette for optional args — namely an **initial prompt** (agent presets only), which patterm injects into the spawned PTY's input after the agent is ready (§8). The orchestrator equivalent of this — `spawn_agent` / `spawn_process` MCP tools — uses the exact same machinery: pick a preset by name, optionally supply an initial prompt, patterm handles the rest.
 
@@ -365,11 +365,11 @@ Risks acknowledged: the orchestrator's reading of the prompt is a vision/parsing
 
 ## 10. Presets
 
-Presets are user-editable JSON files that describe how to launch something. patterm itself has no hard-coded agent or process types — every spawnable thing is a preset. Two flavours:
+Presets describe how to launch something. patterm has built-in defaults for common agent CLIs, and user-editable JSON files can override, disable, or add presets. Two flavours:
 
 ### Agent presets
 
-`$XDG_CONFIG_HOME/patterm/presets/agents/<name>.json`. Launches a vendor LLM CLI with MCP wired up and the conversation-protocol addendum injected.
+Built-in agent presets launch vendor LLM CLIs with MCP wired up and the conversation-protocol addendum injected. `$XDG_CONFIG_HOME/patterm/presets/agents/<name>.json` can overlay a built-in by `name` or define a new agent preset.
 
 | Field | Purpose |
 |---|---|
@@ -377,17 +377,18 @@ Presets are user-editable JSON files that describe how to launch something. patt
 | `argv` | Full launch argv (e.g. `["claude"]`, `["codex", "--no-tui-banner"]`) |
 | `env` | Env vars to set (merged over inherited env) |
 | `working_dir` | Defaults to the project root |
+| `disabled` | If `true`, hides a built-in preset with the same `name` |
 | `mcp_injection` | How to point this CLI at patterm's stdio proxy. One of: `{ "kind": "flag", "flag": "--mcp-config", "config_path": "..." }`, `{ "kind": "config_file", "path": "~/.codex/config.toml", "merge_key": "mcp_servers" }`, `{ "kind": "env_var", "var": "MCP_CONFIG_PATH" }` |
 | `ready_signal` | How to detect the TUI is ready (default: 1s idle after launch). Override per-CLI if needed. |
 | `chrome_trim_hints` | Optional regexes / row ranges for stripping vendor chrome in grid reads |
 
-Default presets shipped: `claude`, `codex`, `opencode`. Authoring these is per-vendor research — each CLI has its own MCP config conventions, ready states, and TUI chrome. Users can copy and edit them, or add new ones (e.g. a second `claude` preset that launches with a specific model or system prompt file).
+Built-in presets: `claude`, `codex`, `opencode`. Authoring these is per-vendor research — each CLI has its own MCP config conventions, ready states, idle detection, and TUI chrome. Users can add small overlay files for built-ins, disable built-ins, or add new presets (e.g. a second `claude-sonnet` preset that launches with a specific model or system prompt file).
 
 MCP config flow: at startup, for each agent preset, patterm renders a small JSON pointing at its own `mcp-stdio` proxy subcommand (`patterm mcp-stdio --socket <pid-sock> --identity <token>`) into a per-preset temp file. The launch then uses the preset's `mcp_injection` strategy to hand that path to the CLI. The user's global vendor config is never mutated.
 
 ### Process presets
 
-`$XDG_CONFIG_HOME/patterm/presets/processes/<name>.json`. Launches a raw command in a PTY — no MCP, no addendum, no system prompt.
+`$XDG_CONFIG_HOME/patterm/presets/processes/<name>.json`. Launches a raw command in a PTY — no MCP, no addendum, no system prompt. There are no built-in process presets.
 
 | Field | Purpose |
 |---|---|

TODO.md

@@ -1,7 +0,0 @@
-- [ ] There's a unicode <?> being displayed in opencode [ON HOLD]
-  - Investigated 2026-05-14: patterm passes ghostty grapheme codepoints
-    through unchanged (vt/ghostty.go:452-462), so the `<?>` glyph is
-    most likely the *host* terminal's font fallback for opencode's
-    Nerd Font private-use codepoints, not a patterm substitution.
-    Need a concrete reproduction (which codepoint, which host
-    terminal/font) before changing rendering.

cmd/patterm/debug_harness.go

@@ -2,10 +2,11 @@ package main
 
 import (
 	"encoding/json"
-	"flag"
 	"fmt"
 	"os"
 
+	flag "github.com/spf13/pflag"
+
 	"github.com/hjbdev/patterm/internal/harness"
 )
 

cmd/patterm/main.go

@@ -4,6 +4,7 @@
 //
 //	patterm                      run in $PWD
 //	patterm --project <dir>      run in <dir>
+//	patterm --version            print version and exit
 //	patterm mcp-stdio --socket S --identity I
 //	                             internal: stdio MCP proxy spawned for
 //	                             children, forwards JSON-RPC over S
@@ -13,15 +14,28 @@ package main
 
 import (
 	"context"
-	"flag"
+	"encoding/json"
 	"fmt"
+	"net"
 	"os"
+	"path/filepath"
+	"runtime"
+	"runtime/debug"
+	"runtime/pprof"
+	"time"
+
+	flag "github.com/spf13/pflag"
 
 	"github.com/hjbdev/patterm/internal/app"
 	"github.com/hjbdev/patterm/internal/mcp"
 	"github.com/hjbdev/patterm/internal/projectkey"
+	"github.com/hjbdev/patterm/internal/protocol"
 )
 
+// version is overridden at build time via `-ldflags "-X main.version=..."`.
+// Defaults to "dev" so source builds are still meaningful.
+var version = "dev"
+
 func main() {
 	// The mcp-stdio subcommand is a separate top-level mode: when an
 	// agent CLI launches `patterm mcp-stdio --socket ...`, the same
@@ -37,16 +51,47 @@ func main() {
 		runDebugHarness()
 		return
 	}
+	if len(os.Args) >= 2 && os.Args[1] == "daemon" {
+		os.Args = append(os.Args[:1], os.Args[2:]...)
+		runDaemonCommand()
+		return
+	}
+	if len(os.Args) >= 2 && os.Args[1] == "connect" {
+		os.Args = append(os.Args[:1], os.Args[2:]...)
+		runConnectCommand()
+		return
+	}
+	if len(os.Args) >= 2 && os.Args[1] == "ls" {
+		runDaemonList()
+		return
+	}
 
-	var projectDir = flag.String("project", "", "project directory (default $PWD)")
+	var (
+		projectDir  = flag.String("project", "", "project directory (default $PWD)")
+		showVersion = flag.Bool("version", false, "print version and exit")
+		inProcess   = flag.Bool("in-process", false, "run the legacy single-process TUI instead of attaching to the daemon")
+		debugDir    = flag.String("debug", "", "write debug logs + per-child raw PTY output to DIR (auto-picks a dated subdir under $XDG_STATE_HOME/patterm/debug when DIR is omitted)")
+		profileDir  = flag.String("profile", "", "write pprof files (cpu/heap/goroutine) and live perf counters (metrics.jsonl per-second, metrics.json + summary.txt on exit) to DIR (auto-picks a dated subdir under $XDG_STATE_HOME/patterm/profile when DIR is omitted)")
+	)
+	// Allow bare `--debug` / `--profile` with no value — pflag treats
+	// them as boolean-shaped strings, picking a sensible default dir.
+	flag.Lookup("debug").NoOptDefVal = "auto"
+	flag.Lookup("profile").NoOptDefVal = "auto"
 	flag.Parse()
 
+	if *showVersion {
+		fmt.Println(versionString())
+		return
+	}
+
 	cwd, err := os.Getwd()
 	if err != nil {
 		die("getwd: %v", err)
 	}
 	if *projectDir != "" {
 		cwd = *projectDir
+	} else if flag.NArg() > 0 {
+		cwd = flag.Arg(0)
 	}
 	key, err := projectkey.Key(cwd)
 	if err != nil {
@@ -57,15 +102,119 @@ func main() {
 		die("chdir %s: %v", cwd, err)
 	}
 
+	resolvedDebug, err := resolveDiagDir(*debugDir, "debug")
+	if err != nil {
+		die("debug: %v", err)
+	}
+	resolvedProfile, err := resolveDiagDir(*profileDir, "profile")
+	if err != nil {
+		die("profile: %v", err)
+	}
+
+	stopProfile := startProfile(resolvedProfile)
+	defer stopProfile()
+
 	ctx := context.Background()
-	if err := app.Run(ctx, app.Options{
+	if *inProcess || os.Getenv("PATTERM_NO_DAEMON") != "" {
+		if err := app.Run(ctx, app.Options{
+			ProjectDir: cwd,
+			ProjectKey: key,
+			DebugDir:   resolvedDebug,
+			ProfileDir: resolvedProfile,
+		}); err != nil {
+			die("%v", err)
+		}
+		return
+	}
+	if resolvedDebug != "" || resolvedProfile != "" {
+		die("--debug and --profile currently require --in-process")
+	}
+	if err := app.RunAttachedClient(ctx, app.ClientOptions{
 		ProjectDir: cwd,
-		ProjectKey: key,
+		Stdin:      os.Stdin,
+		Stdout:     os.Stdout,
+		RawMode:    true,
+		AutoStart:  true,
 	}); err != nil {
 		die("%v", err)
 	}
 }
 
+// resolveDiagDir turns the raw flag value into an absolute directory
+// path. Empty string disables the feature. The sentinel "auto" (set by
+// NoOptDefVal on bare flags) picks $XDG_STATE_HOME/patterm/<kind>/<ts>.
+// Any other value is treated as a literal path.
+func resolveDiagDir(raw, kind string) (string, error) {
+	if raw == "" {
+		return "", nil
+	}
+	if raw == "auto" {
+		base := os.Getenv("XDG_STATE_HOME")
+		if base == "" {
+			home, err := os.UserHomeDir()
+			if err != nil {
+				return "", err
+			}
+			base = filepath.Join(home, ".local", "state")
+		}
+		ts := time.Now().Format("20060102-150405")
+		return filepath.Join(base, "patterm", kind, ts), nil
+	}
+	return raw, nil
+}
+
+// startProfile begins a CPU profile under dir and returns a stop func
+// that writes heap + goroutine snapshots before flushing the CPU file.
+// Returns a no-op stop func when dir is empty. All diagnostics are
+// written to <dir>/profile.log — never to stdout/stderr — so the TUI
+// stays uncluttered.
+func startProfile(dir string) func() {
+	if dir == "" {
+		return func() {}
+	}
+	if err := os.MkdirAll(dir, 0o700); err != nil {
+		return func() {}
+	}
+	logPath := filepath.Join(dir, "profile.log")
+	plog := func(format string, args ...any) {
+		f, err := os.OpenFile(logPath, os.O_CREATE|os.O_WRONLY|os.O_APPEND, 0o600)
+		if err != nil {
+			return
+		}
+		defer f.Close()
+		fmt.Fprintf(f, format+"\n", args...)
+	}
+	cpuPath := filepath.Join(dir, "cpu.pprof")
+	f, err := os.Create(cpuPath)
+	if err != nil {
+		plog("cpu open: %v", err)
+		return func() {}
+	}
+	if err := pprof.StartCPUProfile(f); err != nil {
+		plog("cpu start: %v", err)
+		_ = f.Close()
+		return func() {}
+	}
+	plog("profiling started at %s", time.Now().Format(time.RFC3339Nano))
+	return func() {
+		pprof.StopCPUProfile()
+		_ = f.Close()
+		// Heap and goroutine snapshots at exit. Heap captures
+		// steady-state allocation; goroutine catches stragglers
+		// that didn't get cleaned up.
+		runtime.GC()
+		if hf, err := os.Create(filepath.Join(dir, "heap.pprof")); err == nil {
+			_ = pprof.Lookup("heap").WriteTo(hf, 0)
+			_ = hf.Close()
+		}
+		if gf, err := os.Create(filepath.Join(dir, "goroutine.pprof")); err == nil {
+			_ = pprof.Lookup("goroutine").WriteTo(gf, 0)
+			_ = gf.Close()
+		}
+		plog("profiling stopped at %s", time.Now().Format(time.RFC3339Nano))
+	}
+}
+
 func runMCPProxy() {
 	var (
 		socket   = flag.String("socket", "", "path to the running patterm process's MCP socket")
@@ -80,6 +229,168 @@ func runMCPProxy() {
 	}
 }
 
+func runDaemonCommand() {
+	if len(os.Args) >= 2 && os.Args[1] == "stop" {
+		runDaemonStop()
+		return
+	}
+	if len(os.Args) >= 2 && os.Args[1] == "ls" {
+		runDaemonList()
+		return
+	}
+	var (
+		projectDir = flag.String("project", "", "initial project directory (default $PWD)")
+		listenAddr = flag.String("listen", "", "optional TCP listen address for remote human clients (for example 127.0.0.1:2488, 0.0.0.0:2488, or 2488)")
+	)
+	flag.Parse()
+	cwd, err := os.Getwd()
+	if err != nil {
+		die("getwd: %v", err)
+	}
+	if *projectDir != "" {
+		cwd = *projectDir
+	} else if flag.NArg() > 0 {
+		cwd = flag.Arg(0)
+	}
+	if err := app.RunDaemon(context.Background(), app.DaemonOptions{ProjectDir: cwd, ListenAddr: *listenAddr}); err != nil {
+		die("daemon: %v", err)
+	}
+}
+
+func runConnectCommand() {
+	var (
+		host       = flag.String("host", "", "remote daemon host:port")
+		token      = flag.String("token", "", "remote daemon token (default PATTERM_TOKEN or stored token file)")
+		projectDir = flag.String("project", "", "project directory to request on the daemon")
+	)
+	flag.Parse()
+	if *host == "" && flag.NArg() > 0 {
+		*host = flag.Arg(0)
+	}
+	if *host == "" {
+		die("connect: --host HOST:PORT is required")
+	}
+	tok := *token
+	if tok == "" {
+		tok = os.Getenv("PATTERM_TOKEN")
+	}
+	if tok == "" {
+		if stored, err := app.LoadClientToken(); err == nil {
+			tok = stored
+		}
+	}
+	if tok == "" {
+		die("connect: token required via --token, PATTERM_TOKEN, or %s", mustTokenPath())
+	}
+	cwd := *projectDir
+	if cwd == "" {
+		var err error
+		cwd, err = os.Getwd()
+		if err != nil {
+			die("getwd: %v", err)
+		}
+	}
+	tr, err := app.DialTCPTransport(*host)
+	if err != nil {
+		die("connect: %v", err)
+	}
+	defer tr.Close()
+	if err := app.RunAttachedClient(context.Background(), app.ClientOptions{
+		ProjectDir: cwd,
+		Transport:  tr,
+		Stdin:      os.Stdin,
+		Stdout:     os.Stdout,
+		RawMode:    true,
+		Token:      tok,
+	}); err != nil {
+		die("connect: %v", err)
+	}
+}
+
+func mustTokenPath() string {
+	path, err := app.ClientTokenPath()
+	if err != nil {
+		return "$XDG_DATA_HOME/patterm/clients/token"
+	}
+	return path
+}
+
+func runDaemonList() {
+	projects, err := daemonRequest(protocol.Frame{Type: protocol.FrameList})
+	if err != nil {
+		die("ls: %v", err)
+	}
+	for _, p := range projects.Projects {
+		fmt.Printf("%s\t%d\t%s\n", p.Key, p.TabCount, p.Path)
+	}
+}
+
+func runDaemonStop() {
+	if _, err := daemonRequest(protocol.Frame{Type: protocol.FrameStop}); err != nil {
+		die("daemon stop: %v", err)
+	}
+	fmt.Println("stopped")
+}
+
+func daemonRequest(req protocol.Frame) (protocol.ProjectList, error) {
+	socket, _, err := app.RuntimeDaemonPaths()
+	if err != nil {
+		return protocol.ProjectList{}, err
+	}
+	conn, err := net.Dial("unix", socket)
+	if err != nil {
+		return protocol.ProjectList{}, err
+	}
+	defer conn.Close()
+	t := protocol.NewConnTransport(conn)
+	if err := t.Send(req); err != nil {
+		return protocol.ProjectList{}, err
+	}
+	resp, err := t.Recv()
+	if err != nil {
+		return protocol.ProjectList{}, err
+	}
+	if resp.Type == protocol.FrameError {
+		var msg protocol.Error
+		_ = json.Unmarshal(resp.Payload, &msg)
+		if msg.Message == "" {
+			msg.Message = "daemon returned an error"
+		}
+		return protocol.ProjectList{}, fmt.Errorf("%s", msg.Message)
+	}
+	if resp.Type != protocol.FrameProjectList {
+		return protocol.ProjectList{}, fmt.Errorf("unexpected daemon response %q", resp.Type)
+	}
+	return protocol.Decode[protocol.ProjectList](resp)
+}
+
+func versionString() string {
+	commit, date := "unknown", "unknown"
+	if info, ok := debug.ReadBuildInfo(); ok {
+		dirty := false
+		for _, s := range info.Settings {
+			switch s.Key {
+			case "vcs.revision":
+				if len(s.Value) >= 7 {
+					commit = s.Value[:7]
+				} else if s.Value != "" {
+					commit = s.Value
+				}
+			case "vcs.time":
+				if t, err := time.Parse(time.RFC3339, s.Value); err == nil {
+					date = t.Format("2006-01-02")
+				}
+			case "vcs.modified":
+				dirty = s.Value == "true"
+			}
+		}
+		if dirty && commit != "unknown" {
+			commit += "-dirty"
+		}
+	}
+	return fmt.Sprintf("patterm %s (commit %s, built %s)", version, commit, date)
+}
+
 func die(format string, args ...any) {
 	fmt.Fprintf(os.Stderr, "patterm: "+format+"\n", args...)
 	os.Exit(1)

cmd/spike/main.go

@@ -108,7 +108,7 @@ func run(argv []string, cols, rows uint16, idleMS int, followHost, stdinPassthro
 	}
 	defer em.Close()
 
-	child, err := pty.Start(argv, nil, cols, rows)
+	child, err := pty.Start(argv, nil, "", cols, rows)
 	if err != nil {
 		return fmt.Errorf("pty: %w", err)
 	}

docs/daemon-client-plan.md

@@ -0,0 +1,273 @@
+# patterm: persistent daemon + thin networked client — implementation plan
+
+Status: implemented — Phases 0–4 landed on this branch. Branch: `feat/daemon-client-split`.
+
+> Implemented: pty workdir/process-group + protocol/Transport/loopback foundation;
+> multi-project `ProjectRegistry`; out-of-process unix-socket daemon with auto-start,
+> `daemon stop`/`ls`, detach (Ctrl-]) + reconnect; opt-in LAN TCP listener with a
+> lightweight bearer token + `patterm connect`; per-pane display-owner sizing for
+> multi-client viewing. Deferred (not built): TLS (transport kept pluggable),
+> remote MCP, durable restore of live PTYs across daemon restart.
+
+## Goal
+
+Turn patterm from a single foreground process into a persistent background
+**daemon** that owns all process/project state, plus a thin **client** that
+renders and forwards input. A client on another LAN device can attach,
+navigate projects via the command palette, detach, and reconnect — with child
+processes surviving across client disconnects.
+
+## Locked decisions
+
+1. **Scope:** build all phases; land as one PR off this branch.
+2. **Remote access:** human UI clients only. MCP for agents stays local
+   (per-daemon unix socket); no remote MCP transport in this work.
+3. **Multi-client = per-client independent view.** The daemon holds pure
+   process/project state. Each client connection owns a `ClientView`
+   (selected project, focused pane/pad, scroll offset, palette state,
+   terminal size). Two clients may sit on different projects at once.
+4. **Daemon lifecycle:** auto-start on demand (tmux/docker model). `patterm`
+   starts the daemon if absent and attaches; `patterm daemon stop|ls` manage it.
+5. **Durability:** "persistent" = survive client disconnect while the daemon
+   process lives. Daemon restart only rehydrates today's persist model
+   (top-level commands, fresh IDs). No attempt to resurrect live PTYs/agents
+   after daemon death.
+6. **Auth (trusted-network stance):** Harry runs this on a trusted LAN and is
+   fine with LAN exposure. Keep it lightweight: localhost default, opt-in LAN
+   bind (`--listen`), a simple pairing/bearer token to prevent accidental
+   drive-by access. TLS/cert-pinning is NOT required now but the transport must
+   stay pluggable so TLS can be layered in later.
+7. **Detach gesture:** explicit detach via a palette command and/or a dedicated
+   host chord. Ctrl-D stays as PTY input (shell EOF), as today. Quit-project and
+   stop-daemon are explicit actions.
+
+## Current architecture (baseline facts — verify before editing)
+
+- `app.Run` (`internal/app/app.go:49`) wires the entire process: presets,
+  settings, scratchpad/trust/persist stores, in-process MCP server, ONE
+  `Session`, the `uiState` TUI, classifier, SIGWINCH, 60Hz chrome ticker,
+  blocking `stdinLoop`.
+- **The seam:** `ChildEventListener` (`internal/app/session.go:83`) —
+  `OnChildSpawned`/`OnChildExited`/`OnPTYOut`/`OnChildStateChanged`/
+  `OnChildClosed`. Today `uiState` is the only real listener (subscribed at
+  `app.go:198`). A remote client = a serialized listener + reverse command
+  channel.
+- One `Session` (`session.go:28`) holds a flat `children map[string]*Child` +
+  `order`. Tabs are derived: `KindAgent` children with `ParentID==""`
+  (`tree.go` `runningTopLevels`). The whole tree is reconstructed from
+  `Child.ParentID`.
+- `Child` (`child.go:72`) owns `*pty.PTY`, `*vt.GhosttyEmulator`, raw ring,
+  status/owner atomics. Lifecycle: `Session.Spawn` (`session.go:222`) →
+  `startPTY` → `pumpChild` (`session.go:423`, PTY→emulator→ring→`emitPTYOut`)
+  + `reapChild` (`session.go:488`, exit→`killDescendantsOf`).
+- Stores already keyed by projectKey on `Open`
+  (`scratchpad`/`trust`/`persist`); `projectkey.Key(dir)` =
+  `sha256(realpath)[:16]`.
+- `SerializeChild` (`session.go:687`) already yields a full VT snapshot for
+  stateless repaint.
+- Rendering writes ANSI to `os.Stdout` under `outMu`; `viewportRenderer`
+  (`internal/app/viewport_renderer.go`) is a stateful ANSI rewriter confining
+  child output to the viewport. Input: raw `os.Stdin` via `stdinLoop`
+  (`app.go:1433`)/`processStdin`.
+- MCP: in-process `Server` (`internal/mcp/mcp.go:26`), newline-JSON over a
+  per-PID unix socket `$XDG_RUNTIME_DIR/patterm/<pid>.sock`. Agents launch
+  `patterm mcp-stdio --socket S --identity T`. Identity → `callerID` via
+  `host.ResolveCallerIdentity` → `Session.FindChildByIdentity`.
+- **No TCP/TLS anywhere today.** All `net.Listen`/`net.Dial` are unix sockets.
+- **Must-fix:** `pty.Start` (`internal/pty/pty.go:26`) does not set `cmd.Dir`;
+  today the process `os.Chdir`s once. A daemon can't chdir globally, so
+  `SpawnSpec.WorkDir` must propagate to `exec.Cmd.Dir`.
+
+## Target component model
+
+| Component | Owns |
+|---|---|
+| `internal/daemon` (`pattermd`) | Project registry (N `Session`s), all PTYs, emulators, MCP server, per-project stores, classifier, timers. No TTY. |
+| `internal/client` (`patterm`) | Real terminal: raw mode, alt-screen, SIGWINCH, stdin/stdout; `uiState`, `viewportRenderer`, chrome draws, palette, input. Holds `ClientView`. |
+| `internal/transport` | `Transport` interface + framing; loopback, unix, TCP/TLS impls; auth handshake. |
+| `internal/protocol` | Wire message types shared by daemon + client. |
+
+### `Transport` interface (migration linchpin)
+
+```go
+type Transport interface {
+    Send(Frame) error          // client→daemon command, or daemon→client push
+    Recv() (Frame, error)
+    Close() error
+}
+```
+
+- **Loopback impl:** in-process channels, zero serialization. Default
+  `patterm` = client + loopback daemon in one process → today's UX preserved
+  exactly, single binary.
+- **Net impl:** framed JSON-per-line over `net.Conn`, reusing the
+  `mcp.go:handleConn` pattern; unix socket first, then TCP/TLS.
+
+### Per-client state vs daemon state
+
+```go
+// daemon-side, pure process/project state
+type Registry struct { projects map[string]*Project }  // key = projectKey
+type Project struct {
+    Key, Dir, Name string
+    Session  *Session
+    Pads     *scratchpad.Store
+    Trust    *trust.Store
+    Persist  *persist.Store
+    Launcher *Launcher
+    Host     *ToolHost
+}
+
+// per-connection, client-owned view state (lives client-side; daemon tracks
+// only what it must to size emulators + route subscriptions)
+type ClientView struct {
+    ID          string
+    ProjectKey  string   // which project this client is looking at
+    FocusedID   string   // pane (Child) or pad
+    ScrollOff   int
+    Cols, Rows  uint16
+    // palette state is fully client-local
+}
+```
+
+Project switch = re-point this client's subscription to another `Project`'s
+Session + send `chrome` + `pane_snapshot`. No process teardown.
+
+### Wire protocol (control + UI channel)
+
+Bidirectional framed JSON-per-line.
+
+Daemon → client:
+- `hello` / `auth_challenge` / `auth_ok` — handshake.
+- `project_list` — `[{key, path, name, last_active, tab_count}]` for the
+  palette switcher.
+- `chrome` — semantic model for the client's current project+view: tab list
+  (`runningTopLevels`), sidebar tree (`sidebarNav`), status/owner, toasts,
+  scratchpad list + selected preview. Client draws chrome locally
+  (reuses `tabbar.go`/`sidebar.go`).
+- `pane_snapshot{paneID, vtBytes}` — full repaint on focus/attach/switch via
+  `SerializeChild`.
+- `pane_chunk{paneID, bytes}` — live focused-pane PTY output (serialized
+  `OnPTYOut`).
+- `lifecycle{spawned|exited|closed|stateChanged,...}` — serialized listener.
+- `attention` / `trust_prompt` — human-facing surfaces; render on the client
+  whose view owns the relevant project.
+
+Client → daemon:
+- `attach{token, term_size, project_key?}` / `detach`.
+- `input{paneID, bytes}` (the `InjectAsUser` path).
+- `focus{paneID|pad}`, `switch_project{key}`, `open_project{path}`.
+- `palette_command{...}` (spawn/kill/rename/quit-project), `trust_response`,
+  `resize{cols,rows}`.
+
+**Encoding decision:** ship raw focused-pane PTY bytes + periodic
+`SerializeChild` snapshots; client runs its own `viewportRenderer`. No
+daemon-side pre-render (keeps daemon size-agnostic), no grid diffs in v1.
+Requires in-order delivery only (TCP gives it). Diffs are a later optimization.
+
+### Emulator sizing with per-client views
+
+Each `Child` emulator has one size. Rules:
+- A pane is sized by the client(s) viewing it. If exactly one client focuses a
+  pane, that client's cols/rows drive `ResizeAll` for that pane.
+- If two clients focus the **same** pane, one is the **display owner** (first
+  to focus, or explicit take-control); the owner's size drives the emulator;
+  the other letterboxes/clips. Surface a toast.
+- Because clients are usually on different projects/panes, contention is rare.
+
+### Security (human clients, LAN — trusted-network stance)
+
+Harry runs this on a trusted LAN (decision #6). Keep it lightweight but not
+wide open:
+- localhost-only by default. LAN bind (`--listen 0.0.0.0:PORT`) is explicit
+  opt-in, never default.
+- A simple pairing/bearer token gates network attach so a stray host on the LAN
+  can't drive-by-attach. Daemon prints the token on `--listen`; client presents
+  it in `attach`; store a per-client token after first pairing.
+- Local unix-socket clients keep `0600` perms (sufficient for same-user).
+- Keep the transport pluggable so TLS + cert pinning can be layered in later
+  without reworking the protocol. Not building TLS now.
+- Trust prompts may now be approved from another device — deliberate; route to
+  the client whose view owns the project.
+
+### Daemon lifecycle (auto-start)
+
+- Well-known local socket `$XDG_RUNTIME_DIR/patterm/daemon.sock` +
+  pidfile/lockfile (single daemon per user).
+- `patterm [dir]`: dial the socket; if absent, fork-exec the daemon, wait for
+  readiness, attach. `--project`/dir selects the initial project for the view.
+- `patterm daemon` (foreground), `patterm daemon stop`, `patterm ls`.
+- **Detach = explicit** palette command and/or a dedicated host chord; PTYs keep
+  running. Ctrl-D stays as PTY input (shell EOF). Quitting a project / killing
+  the daemon are explicit palette/CLI actions.
+- Idle-shutdown policy: configurable; default keep alive until explicit stop.
+
+## Package-by-package changes
+
+- **`cmd/patterm`** (`main.go`): add `daemon` subcommand (headless core);
+  default invocation becomes client (auto-start/attach); `mcp-stdio` dials the
+  shared daemon socket (not per-PID); `debug-harness` drives a daemon (or
+  loopback).
+- **`internal/app` split:**
+  - new **`internal/daemon`**: headless half — move `session.go`, `child.go`,
+    `host.go`, `tree.go`, `launch.go`, classifier, timers, `Shutdown`,
+    kill-cascade. Add `Registry`/`Project`.
+  - **`internal/client`**: TTY half — `uiState`, `viewport_renderer.go`,
+    `screen_renderer.go`, `tabbar.go`, `sidebar.go`, status, `palette.go`,
+    `stdinLoop`/`processStdin`, SIGWINCH/chrome ticker, markdown/marquee/toast.
+    Consumes events + chrome over `Transport` instead of `sess.Subscribe`.
+- **new `internal/transport` + `internal/protocol`**: messages, framing,
+  loopback/unix/TCP-TLS impls, auth handshake.
+- **`internal/mcp`**: `SocketPath` per-daemon (not per-PID);
+  `ResolveCallerIdentity` becomes daemon-wide across projects (token already
+  carries `PATTERM_PROJECT_KEY` via `ChildEnv`).
+- **`internal/pty`**: set `cmd.Dir` from `SpawnSpec.WorkDir`; add process-group
+  handling for reliable tree teardown.
+- **`internal/vt`**: unchanged grid source of truth; enforce per-child
+  serialization around emulator access (interface isn't concurrency-safe) since
+  clients + MCP + pump all snapshot.
+- **`internal/{scratchpad,trust,persist}`**: per-`Project` instances in the
+  registry (already keyed by projectKey).
+- **`internal/preset`**: project-agnostic; daemon loads once, shares.
+- **`internal/projectkey`**: doc update (key is now load-bearing for routing).
+- **`internal/harness`**: add daemon/loopback mode; assert child survives client
+  disconnect/reconnect, project-switch preserves each project's tree, two
+  clients on different projects, unauth TCP rejected.
+
+## Backpressure
+
+`pumpChild`'s listener calls are synchronous (`session.go:149`). A slow network
+client must not block the PTY pump. Introduce a per-client event bus with a
+bounded buffer that coalesces/ drops to a snapshot under pressure, decoupled
+from `pumpChild`.
+
+## Phased roadmap (all phases land on this branch)
+
+0. **Extract headless core behind loopback transport.** `daemon.Core` +
+   `client` over in-process `Transport`. Zero behavior change; harness green.
+1. **Multi-project registry + per-client view scaffolding.** Registry, per-
+   project stores, `ClientView`, palette "Switch/Open project…", project tier
+   in chrome. Still single local process.
+2. **Out-of-process daemon over unix socket.** Auto-start/attach; PTYs survive
+   client exit; reconnect + snapshot-on-attach; Ctrl-D = detach; pidfile/lock.
+3. **TCP + TLS + auth.** localhost TCP, then opt-in LAN bind; pairing token /
+   cert pinning; remote trust-prompt routing.
+4. **Per-client view fully realized + emulator sizing/display-owner.**
+   Independent focus/scroll/palette per client; multi-client on same/different
+   projects; resize negotiation + letterbox.
+5. **Hardening.** systemd/launchd autostart, `daemon stop|ls`, idle-shutdown,
+   backpressure, security review, CHANGELOG.
+
+## Risks / open questions for review
+
+- Heterogeneous client sizes vs one-PTY-one-size (display-owner + letterbox is
+  the v1 answer — is it sufficient?).
+- Security escalation: a network client spawns processes / runs shell / injects
+  input. Auth/TLS scope adequate?
+- Ctrl-D semantics flip — acceptable UX?
+- Backpressure design — bounded bus + snapshot-on-pressure correct?
+- MCP identity uniqueness across projects after per-PID socket removal.
+- Is per-client view (decision #3) worth doing from Phase 1, or staged after a
+  shared-focus interim that's faster to ship?
+- Splitting `uiState` (focus/palette/render caches/trust prompt/dims/outMu) out
+  of the daemon is the largest refactor — sequencing concerns?

go.mod

@@ -4,6 +4,7 @@ go 1.26.3
 
 require (
 	github.com/creack/pty v1.1.24
+	github.com/spf13/pflag v1.0.10
 	golang.org/x/term v0.43.0
 )
 

go.sum

@@ -1,5 +1,7 @@
 github.com/creack/pty v1.1.24 h1:bJrF4RRfyJnbTJqzRLHzcGaZK1NeM5kTC9jGgovnR1s=
 github.com/creack/pty v1.1.24/go.mod h1:08sCNb52WyoAwi2QDyzUCTgcvVFhUzewun7wtTfvcwE=
+github.com/spf13/pflag v1.0.10 h1:4EBh2KAYBwaONj6b2Ye1GiHfwjqyROoF4RwYO+vPwFk=
+github.com/spf13/pflag v1.0.10/go.mod h1:McXfInJRrz4CZXVZOBLb0bTZqETkiAhM9Iw0y3An2Bg=
 golang.org/x/sys v0.44.0 h1:ildZl3J4uzeKP07r2F++Op7E9B29JRUy+a27EibtBTQ=
 golang.org/x/sys v0.44.0/go.mod h1:4GL1E5IUh+htKOUEOaiffhrAeqysfVGipDYzABqnCmw=
 golang.org/x/term v0.43.0 h1:S4RLU2sB31O/NCl+zFN9Aru9A/Cq2aqKpTZJ6B+DwT4=

idle-detection.md

@@ -0,0 +1,26 @@
+# Idle Detection
+
+Solo does idle detection to show which agents are running, but this can also allow sub-agents to read state and/or trigger timers/actions based on idle state. This is important for things like permission checks. If an agent becomes idle, the orchestrator needs to know so it can approve permissions etc.
+
+<solo-idle-detection-docs>
+Agent idle detection
+Solo tracks agent state so you can tell which agents are working, idle, waiting for permission, or blocked by an error.
+
+How it works#
+Solo uses a mix of signals:
+
+First-party terminal agents use provider-specific activity strategies. Claude and OpenCode use visible output, Codex and Amp use OSC title stability, and Gemini uses OSC title status.
+Auto-summarization can return one of IDLE, PERMISSION, THINKING, WORKING, or ERROR, and Solo stores that classification when available.
+Summary timing#
+For summaries, Solo waits until a process has had human input and then watches output activity. A brief quiet window can trigger a summary after output stops. Continuously busy processes can also trigger summaries after a longer busy window.
+
+The summary cadence setting is still enforced per process, so repeated activity does not produce unlimited summary attempts.
+
+Timers#
+Agents can also have timers through Solo's agent-channel tools. Timer indicators show the nearest active or paused timer on the process row. Clicking the timer lets you view its message, cancel it, fire it now, or pause/resume it.
+
+When a timer fires, Solo delivers the timer message back to the owning process.
+
+Limits#
+Idle detection is a heuristic. Some agents pause between steps before continuing on their own, and a quiet terminal is not always the same thing as completed work.
+</solo-idle-detection-docs>

install.sh

@@ -0,0 +1,8 @@
+#!/usr/bin/bash
+
+echo "Building Patterm"
+./build.sh
+echo "Installing Patterm"
+sudo cp ./bin/patterm /usr/local/bin
+echo "Done"
+echo "Copied ./bin/patterm to /usr/local/bin"

internal/app/bench_test.go

@@ -0,0 +1,546 @@
+package app
+
+import (
+	"fmt"
+	"io"
+	"strings"
+	"testing"
+
+	"github.com/hjbdev/patterm/internal/vt"
+)
+
+// Benchmarks for patterm's hot paths. Run with:
+//
+//	go test -bench=. -benchmem ./internal/app/
+//
+// or target one:
+//
+//	go test -bench=BenchmarkViewportRenderer_PlainASCII -benchmem ./internal/app/
+//
+// The fixtures below model the three workloads we care about most:
+//
+//   - PlainASCII: long-running text output (claude streaming a code
+//     diff, codex outputting a tool result body). Fast-path territory.
+//   - StyledLines: SGR-heavy output (claude/codex chat history with
+//     coloured tokens). State-machine path.
+//   - RatatuiBurst: many short cursor-positioning / SGR transitions in
+//     a tight chunk, matching codex/ratatui's incremental diff
+//     updates.
+//   - SnapshotReplay: full styled-grid replay (focus switch).
+
+// buildPlainASCIIChunk returns a roughly N-byte chunk of pure
+// printable ASCII text with the occasional newline — the cheapest
+// workload, exercises the fast path in viewport_renderer.Render.
+func buildPlainASCIIChunk(n int) []byte {
+	var b strings.Builder
+	b.Grow(n)
+	line := "The quick brown fox jumps over the lazy dog 0123456789 "
+	for b.Len() < n {
+		b.WriteString(line)
+		if b.Len()%80 < len(line) {
+			b.WriteByte('\n')
+		}
+	}
+	return []byte(b.String()[:n])
+}
+
+// buildStyledLinesChunk simulates SGR-heavy output: every word wears
+// a colour, so the renderer breaks out of its fast path on every
+// escape sequence.
+func buildStyledLinesChunk(n int) []byte {
+	var b strings.Builder
+	b.Grow(n)
+	colours := []string{"31", "32", "33", "34", "35", "36"}
+	words := []string{"package", "func", "return", "import", "struct", "type", "const", "var"}
+	i := 0
+	for b.Len() < n {
+		fmt.Fprintf(&b, "\x1b[%sm%s\x1b[0m ", colours[i%len(colours)], words[i%len(words)])
+		if i%10 == 9 {
+			b.WriteByte('\n')
+		}
+		i++
+	}
+	return []byte(b.String()[:n])
+}
+
+// buildRatatuiBurst simulates a single ratatui-style diff frame:
+// CUP, SGR, a few chars, CUP, SGR, a few chars… for a viewport's
+// worth of cells.
+func buildRatatuiBurst(cells int) []byte {
+	var b strings.Builder
+	for i := 0; i < cells; i++ {
+		row := (i / 80) + 1
+		col := (i % 80) + 1
+		fmt.Fprintf(&b, "\x1b[%d;%dH\x1b[3%dm%c", row, col, i%8, byte('A'+(i%26)))
+	}
+	b.WriteString("\x1b[0m")
+	return []byte(b.String())
+}
+
+// BenchmarkViewportRenderer_PlainASCII drives a 16 KiB plain-text
+// chunk through Render once per iteration. Reports ns/op,
+// allocations, and B/op.
+func BenchmarkViewportRenderer_PlainASCII(b *testing.B) {
+	chunk := buildPlainASCIIChunk(16 * 1024)
+	b.SetBytes(int64(len(chunk)))
+	b.ReportAllocs()
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		vr := newViewportRenderer(newTerminalLayout(120, 40))
+		_ = vr.Render(chunk)
+	}
+}
+
+// BenchmarkViewportRenderer_StyledLines exercises the per-byte CSI
+// path on SGR-heavy output. Most claude/codex chat resume traffic
+// looks like this — coloured prose with frequent style toggles.
+func BenchmarkViewportRenderer_StyledLines(b *testing.B) {
+	chunk := buildStyledLinesChunk(16 * 1024)
+	b.SetBytes(int64(len(chunk)))
+	b.ReportAllocs()
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		vr := newViewportRenderer(newTerminalLayout(120, 40))
+		_ = vr.Render(chunk)
+	}
+}
+
+// BenchmarkViewportRenderer_RatatuiBurst measures the worst-case
+// cursor-shuffling workload: full-frame diff updates dominated by
+// CUP + SGR + single-char writes.
+func BenchmarkViewportRenderer_RatatuiBurst(b *testing.B) {
+	chunk := buildRatatuiBurst(80 * 24) // one screenful of cells
+	b.SetBytes(int64(len(chunk)))
+	b.ReportAllocs()
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		vr := newViewportRenderer(newTerminalLayout(120, 40))
+		_ = vr.Render(chunk)
+	}
+}
+
+// BenchmarkContainsOSC measures the OSC-gate fast path used by
+// pumpChild before deciding whether to fire the per-chunk Title()
+// CGO call. Inputs:
+//   - "hot": SGR-styled output without OSC — the common case for
+//     codex/ratatui. We want this near zero.
+//   - "cold": chunk with an OSC sequence in the middle.
+func BenchmarkContainsOSC_NoOSC(b *testing.B) {
+	chunk := buildStyledLinesChunk(8 * 1024)
+	b.SetBytes(int64(len(chunk)))
+	b.ReportAllocs()
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		_ = containsOSC(chunk)
+	}
+}
+
+func BenchmarkContainsOSC_WithOSC(b *testing.B) {
+	chunk := append(buildStyledLinesChunk(8*1024), []byte("\x1b]0;new title\x07")...)
+	b.SetBytes(int64(len(chunk)))
+	b.ReportAllocs()
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		_ = containsOSC(chunk)
+	}
+}
+
+// BenchmarkRendererThroughput_ReuseInstance approximates real
+// session behaviour: a single viewport renderer fed many chunks in
+// sequence, no per-iteration allocation. Reports a throughput
+// closer to the steady-state OnPTYOut path. Chunks are 4 KiB to
+// match typical PTY read sizes; the renderer is reset every
+// benchmark run.
+func BenchmarkRendererThroughput_ReuseInstance(b *testing.B) {
+	chunks := make([][]byte, 16)
+	for i := range chunks {
+		chunks[i] = buildStyledLinesChunk(4 * 1024)
+	}
+	totalBytes := 0
+	for _, c := range chunks {
+		totalBytes += len(c)
+	}
+	b.SetBytes(int64(totalBytes))
+	b.ReportAllocs()
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		vr := newViewportRenderer(newTerminalLayout(120, 40))
+		for _, c := range chunks {
+			_ = vr.Render(c)
+		}
+	}
+}
+
+// Stress workloads — these model the worst things a real session
+// can throw at us. The headline target is "ASCII video": every cell
+// of an 80x40 viewport carries an SGR colour change and a printable
+// character, rendered as one chunk per frame. Real ASCII-video CLIs
+// (ascii-image-converter, asciinema-render, towel.blinkenlights, the
+// Bad Apple meme) hit patterm with exactly this pattern at 24-30 fps
+// for minutes at a time.
+//
+// We synthesise the workload rather than ship a captured corpus so
+// the benchmarks stay deterministic and the repo doesn't carry tens
+// of MiB of fixture data. The encoding is faithful to what those
+// tools actually emit.
+
+// buildASCIIVideoFrame builds a single full-viewport frame with
+// 8-colour SGR per cell (`\x1b[3Nm`). One frame ≈ 30 KiB for an
+// 80x40 viewport, which lines up with what ascii-video tools emit.
+func buildASCIIVideoFrame(cols, rows int) []byte {
+	var b strings.Builder
+	b.WriteString("\x1b[H") // home cursor before the frame starts
+	for r := 0; r < rows; r++ {
+		for c := 0; c < cols; c++ {
+			fmt.Fprintf(&b, "\x1b[3%dm%c", (r+c)%8, byte(' '+(r*c)%(0x7e-' ')))
+		}
+		b.WriteString("\x1b[0m\r\n")
+	}
+	return []byte(b.String())
+}
+
+// buildASCIIVideoFrameTrueColor builds the same frame but with
+// 24-bit RGB SGR (`\x1b[38;2;R;G;Bm`). Every cell is ~20 bytes of
+// escape + 1 byte glyph, so a frame is ≈ 70 KiB. This is what
+// chafa --colors=full and modern terminal video players emit, and
+// it's the heaviest SGR variant the renderer's CSI path sees.
+func buildASCIIVideoFrameTrueColor(cols, rows int) []byte {
+	var b strings.Builder
+	b.WriteString("\x1b[H")
+	for r := 0; r < rows; r++ {
+		for c := 0; c < cols; c++ {
+			rd := (r * 7) % 256
+			gd := (c * 11) % 256
+			bd := ((r + c) * 13) % 256
+			fmt.Fprintf(&b, "\x1b[38;2;%d;%d;%dm%c", rd, gd, bd, byte(' '+(r*c)%(0x7e-' ')))
+		}
+		b.WriteString("\x1b[0m\r\n")
+	}
+	return []byte(b.String())
+}
+
+// buildBadApplePattern builds the simplest possible ASCII video
+// frame: alternating black/white cells (the Bad Apple meme is
+// essentially a 1-bit silhouette video). This is the pattern that
+// stresses the SGR state-machine without exercising truecolor parse
+// — useful for isolating "is the cost in the colour parsing or in
+// the cell-by-cell switching?"
+func buildBadApplePattern(cols, rows int) []byte {
+	var b strings.Builder
+	b.WriteString("\x1b[H")
+	for r := 0; r < rows; r++ {
+		for c := 0; c < cols; c++ {
+			if (r+c)%2 == 0 {
+				b.WriteString("\x1b[37m█")
+			} else {
+				b.WriteString("\x1b[30m█")
+			}
+		}
+		b.WriteString("\x1b[0m\r\n")
+	}
+	return []byte(b.String())
+}
+
+// BenchmarkASCIIVideo_Frame_8Color renders a single full-screen
+// frame as one chunk. The headline number is MB/s — at 30 fps a
+// frame is one PTY chunk every ~33 ms, so this should comfortably
+// stay well under 1 ms.
+func BenchmarkASCIIVideo_Frame_8Color(b *testing.B) {
+	frame := buildASCIIVideoFrame(80, 40)
+	b.SetBytes(int64(len(frame)))
+	b.ReportAllocs()
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		vr := newViewportRenderer(newTerminalLayout(120, 40))
+		_ = vr.Render(frame)
+	}
+}
+
+// BenchmarkASCIIVideo_Frame_TrueColor renders a single truecolor
+// frame. ~70 KiB per frame. Compare this to the 8-colour number to
+// see how much extra cost the truecolor SGR parse imposes — the
+// `\x1b[38;2;R;G;Bm` form is the longest and most parameter-rich
+// CSI patterm sees in practice.
+func BenchmarkASCIIVideo_Frame_TrueColor(b *testing.B) {
+	frame := buildASCIIVideoFrameTrueColor(80, 40)
+	b.SetBytes(int64(len(frame)))
+	b.ReportAllocs()
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		vr := newViewportRenderer(newTerminalLayout(120, 40))
+		_ = vr.Render(frame)
+	}
+}
+
+// BenchmarkASCIIVideo_Frame_BadApple is the 1-bit pattern: simplest
+// SGR (two colours, alternating). Isolates the renderer's cell-by-
+// cell SGR cycling cost from the truecolor parse cost.
+func BenchmarkASCIIVideo_Frame_BadApple(b *testing.B) {
+	frame := buildBadApplePattern(80, 40)
+	b.SetBytes(int64(len(frame)))
+	b.ReportAllocs()
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		vr := newViewportRenderer(newTerminalLayout(120, 40))
+		_ = vr.Render(frame)
+	}
+}
+
+// runStreamBench is the shared body for the per-fps stream
+// benchmarks. It feeds a fixed frame N times through a single
+// renderer instance and reports µs/frame + an achievable-fps
+// ceiling alongside the standard ns/op + MB/s. The fps value in
+// the benchmark name is the *target* — the workload itself doesn't
+// rate-limit; we just decide how many frames make a benchmark op
+// (3 seconds' worth) so steady-state cost dominates warm-up.
+func runStreamBench(b *testing.B, frame []byte, fps int) {
+	frames := fps * 3 // 3 seconds at the target rate
+	totalBytes := int64(len(frame) * frames)
+	b.SetBytes(totalBytes)
+	b.ReportAllocs()
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		vr := newViewportRenderer(newTerminalLayout(120, 40))
+		for f := 0; f < frames; f++ {
+			_ = vr.Render(frame)
+		}
+	}
+	nsPerFrame := float64(b.Elapsed().Nanoseconds()) / float64(b.N*frames)
+	b.ReportMetric(nsPerFrame/1000.0, "µs/frame")
+	b.ReportMetric(1e9/nsPerFrame, "fps_ceiling")
+	// budget_pct = how much of the per-frame budget at the target
+	// rate we burn. Under 100 means we can hit the target; over
+	// means we can't.
+	budgetNs := 1e9 / float64(fps)
+	b.ReportMetric(nsPerFrame/budgetNs*100, "budget_pct")
+}
+
+// BenchmarkASCIIVideo_Stream_8Color_30fps / _60fps / _120fps reuse
+// one renderer across (3 × fps) frames. The headline numbers are
+// µs/frame, fps_ceiling (= 1e9 / ns/frame), and budget_pct (=
+// percent of the per-frame budget at the target rate we consume).
+//
+// 30 fps is the typical ASCII-video baseline (towel, chafa, Bad
+// Apple ports). 60 is the "smooth playback" target. 120 is a
+// future-proofing stress level matching modern high-refresh
+// terminals.
+func BenchmarkASCIIVideo_Stream_8Color_30fps(b *testing.B) {
+	runStreamBench(b, buildASCIIVideoFrame(80, 40), 30)
+}
+func BenchmarkASCIIVideo_Stream_8Color_60fps(b *testing.B) {
+	runStreamBench(b, buildASCIIVideoFrame(80, 40), 60)
+}
+func BenchmarkASCIIVideo_Stream_8Color_120fps(b *testing.B) {
+	runStreamBench(b, buildASCIIVideoFrame(80, 40), 120)
+}
+
+// BenchmarkASCIIVideo_Stream_TrueColor_* same set but with the
+// truecolor frames. Compare against the 8-colour numbers to see
+// what the longer `\x1b[38;2;R;G;Bm` parse costs us.
+func BenchmarkASCIIVideo_Stream_TrueColor_30fps(b *testing.B) {
+	runStreamBench(b, buildASCIIVideoFrameTrueColor(80, 40), 30)
+}
+func BenchmarkASCIIVideo_Stream_TrueColor_60fps(b *testing.B) {
+	runStreamBench(b, buildASCIIVideoFrameTrueColor(80, 40), 60)
+}
+func BenchmarkASCIIVideo_Stream_TrueColor_120fps(b *testing.B) {
+	runStreamBench(b, buildASCIIVideoFrameTrueColor(80, 40), 120)
+}
+
+// BenchmarkASCIIVideo_Stream_BadApple_* tracks the 1-bit alternating
+// pattern. Isolates per-cell SGR cycling cost from the truecolor
+// parse cost above — useful when reading the diff between the two
+// stream variants.
+func BenchmarkASCIIVideo_Stream_BadApple_30fps(b *testing.B) {
+	runStreamBench(b, buildBadApplePattern(80, 40), 30)
+}
+func BenchmarkASCIIVideo_Stream_BadApple_60fps(b *testing.B) {
+	runStreamBench(b, buildBadApplePattern(80, 40), 60)
+}
+func BenchmarkASCIIVideo_Stream_BadApple_120fps(b *testing.B) {
+	runStreamBench(b, buildBadApplePattern(80, 40), 120)
+}
+
+// BenchmarkEmulator_Write_8Color / _TrueColor isolate the
+// libghostty-vt CGO cost — same frames the Pipeline benchmarks use,
+// but feeding only the emulator. The delta between this and
+// BenchmarkASCIIVideo_Stream_… is the renderer's share; the rest
+// is libghostty-vt.
+func BenchmarkEmulator_Write_8Color_Frame(b *testing.B) {
+	frame := buildASCIIVideoFrame(80, 40)
+	b.SetBytes(int64(len(frame)))
+	b.ReportAllocs()
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		em, err := vt.NewGhosttyEmulator(80, 40)
+		if err != nil {
+			b.Fatalf("emulator: %v", err)
+		}
+		if _, werr := em.Write(frame); werr != nil {
+			b.Fatalf("emulator.Write: %v", werr)
+		}
+		_ = em.Close()
+	}
+}
+
+func BenchmarkEmulator_Write_TrueColor_Frame(b *testing.B) {
+	frame := buildASCIIVideoFrameTrueColor(80, 40)
+	b.SetBytes(int64(len(frame)))
+	b.ReportAllocs()
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		em, err := vt.NewGhosttyEmulator(80, 40)
+		if err != nil {
+			b.Fatalf("emulator: %v", err)
+		}
+		if _, werr := em.Write(frame); werr != nil {
+			b.Fatalf("emulator.Write: %v", werr)
+		}
+		_ = em.Close()
+	}
+}
+
+// BenchmarkEmulator_Write_Stream_120fps reuses one emulator across
+// 360 frames (3 sec × 120 fps). This is the cleanest measurement
+// of em.Write steady-state cost.
+func BenchmarkEmulator_Write_Stream_8Color_120fps(b *testing.B) {
+	frame := buildASCIIVideoFrame(80, 40)
+	const frames = 360
+	b.SetBytes(int64(len(frame) * frames))
+	b.ReportAllocs()
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		em, err := vt.NewGhosttyEmulator(80, 40)
+		if err != nil {
+			b.Fatalf("emulator: %v", err)
+		}
+		for f := 0; f < frames; f++ {
+			if _, werr := em.Write(frame); werr != nil {
+				b.Fatalf("emulator.Write: %v", werr)
+			}
+		}
+		_ = em.Close()
+	}
+	nsPerFrame := float64(b.Elapsed().Nanoseconds()) / float64(b.N*frames)
+	b.ReportMetric(nsPerFrame/1000.0, "µs/frame")
+	b.ReportMetric(1e9/nsPerFrame, "fps_ceiling")
+}
+
+func BenchmarkEmulator_Write_Stream_TrueColor_120fps(b *testing.B) {
+	frame := buildASCIIVideoFrameTrueColor(80, 40)
+	const frames = 360
+	b.SetBytes(int64(len(frame) * frames))
+	b.ReportAllocs()
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		em, err := vt.NewGhosttyEmulator(80, 40)
+		if err != nil {
+			b.Fatalf("emulator: %v", err)
+		}
+		for f := 0; f < frames; f++ {
+			if _, werr := em.Write(frame); werr != nil {
+				b.Fatalf("emulator.Write: %v", werr)
+			}
+		}
+		_ = em.Close()
+	}
+	nsPerFrame := float64(b.Elapsed().Nanoseconds()) / float64(b.N*frames)
+	b.ReportMetric(nsPerFrame/1000.0, "µs/frame")
+	b.ReportMetric(1e9/nsPerFrame, "fps_ceiling")
+}
+
+// runPipelineStreamBench includes the libghostty-vt emulator.Write
+// CGO call and a stdout write to io.Discard alongside the renderer
+// — i.e. everything OnPTYOut does in production except the host
+// terminal's own paint time (which patterm doesn't control). This
+// is the honest "can we hit N fps end-to-end?" measurement.
+func runPipelineStreamBench(b *testing.B, frame []byte, fps int) {
+	frames := fps * 3
+	totalBytes := int64(len(frame) * frames)
+	b.SetBytes(totalBytes)
+	b.ReportAllocs()
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		em, err := vt.NewGhosttyEmulator(80, 40)
+		if err != nil {
+			b.Fatalf("emulator: %v", err)
+		}
+		vr := newViewportRenderer(newTerminalLayout(120, 40))
+		for f := 0; f < frames; f++ {
+			if _, werr := em.Write(frame); werr != nil {
+				b.Fatalf("emulator.Write: %v", werr)
+			}
+			out := vr.Render(frame)
+			// Match OnPTYOut's autowrap prelude/postlude wrapping so
+			// the byte count is faithful.
+			_, _ = io.Discard.Write([]byte("\x1b[?7l"))
+			_, _ = io.Discard.Write(out)
+			_, _ = io.Discard.Write([]byte("\x1b[?7h"))
+		}
+		_ = em.Close()
+	}
+	nsPerFrame := float64(b.Elapsed().Nanoseconds()) / float64(b.N*frames)
+	b.ReportMetric(nsPerFrame/1000.0, "µs/frame")
+	b.ReportMetric(1e9/nsPerFrame, "fps_ceiling")
+	budgetNs := 1e9 / float64(fps)
+	b.ReportMetric(nsPerFrame/budgetNs*100, "budget_pct")
+}
+
+// BenchmarkPipeline_ASCIIVideo_* — the FULL OnPTYOut path
+// (emulator.Write CGO + viewport renderer + a stdout write to
+// io.Discard) running at 30/60/120 fps targets. These are the
+// numbers to trust when asking "can we sustain N fps?" The
+// renderer-only Stream benchmarks above isolate one stage and
+// understate the real cost.
+//
+// 120 fps is the explicit baseline: anything under 100% of the
+// per-frame budget here means we hit 120 fps with margin to spare.
+func BenchmarkPipeline_ASCIIVideo_8Color_30fps(b *testing.B) {
+	runPipelineStreamBench(b, buildASCIIVideoFrame(80, 40), 30)
+}
+func BenchmarkPipeline_ASCIIVideo_8Color_60fps(b *testing.B) {
+	runPipelineStreamBench(b, buildASCIIVideoFrame(80, 40), 60)
+}
+func BenchmarkPipeline_ASCIIVideo_8Color_120fps(b *testing.B) {
+	runPipelineStreamBench(b, buildASCIIVideoFrame(80, 40), 120)
+}
+
+func BenchmarkPipeline_ASCIIVideo_TrueColor_30fps(b *testing.B) {
+	runPipelineStreamBench(b, buildASCIIVideoFrameTrueColor(80, 40), 30)
+}
+func BenchmarkPipeline_ASCIIVideo_TrueColor_60fps(b *testing.B) {
+	runPipelineStreamBench(b, buildASCIIVideoFrameTrueColor(80, 40), 60)
+}
+func BenchmarkPipeline_ASCIIVideo_TrueColor_120fps(b *testing.B) {
+	runPipelineStreamBench(b, buildASCIIVideoFrameTrueColor(80, 40), 120)
+}
+
+// BenchmarkSessionResume_5MiBStyled simulates the user's
+// motivating case: claude resuming a long chat session and dumping
+// the whole history. 5 MiB of styled output as a single Render
+// call. Numbers here tell us how long the visible "scrolling
+// while resume loads" window will be.
+func BenchmarkSessionResume_5MiBStyled(b *testing.B) {
+	chunk := buildStyledLinesChunk(5 * 1024 * 1024)
+	b.SetBytes(int64(len(chunk)))
+	b.ReportAllocs()
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		vr := newViewportRenderer(newTerminalLayout(120, 40))
+		_ = vr.Render(chunk)
+	}
+}
+
+// BenchmarkSessionResume_5MiBPlain same as above but pure text.
+// Lower bound — what we'd hit if the resume content were styling-
+// free.
+func BenchmarkSessionResume_5MiBPlain(b *testing.B) {
+	chunk := buildPlainASCIIChunk(5 * 1024 * 1024)
+	b.SetBytes(int64(len(chunk)))
+	b.ReportAllocs()
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		vr := newViewportRenderer(newTerminalLayout(120, 40))
+		_ = vr.Render(chunk)
+	}
+}

internal/app/child.go

@@ -1,6 +1,7 @@
 package app
 
 import (
+	"bytes"
 	"crypto/rand"
 	"encoding/hex"
 	"errors"
@@ -25,6 +26,11 @@ import (
 // false positives (timestamps, exit codes, etc.).
 var portRegex = regexp.MustCompile(`https?://[^\s:/]+:(\d{2,5})(?:/[^\s]*)?`)
 
+const (
+	agentInterPieceDelay   = 15 * time.Millisecond
+	agentSubmitSettleDelay = 100 * time.Millisecond
+)
+
 type ChildStatus string
 
 const (
@@ -108,16 +114,33 @@ type Child struct {
 
 	// ringMu guards ring. The ring buffer carries the last `ringCap`
 	// bytes the PTY produced, used by SPEC §7 get_process_output stream
-	// mode and search_output scrollback.
+	// mode and search_output scrollback. The ring is a fixed-size byte
+	// array with a wrap-around write index — no per-chunk reslice or
+	// reallocation. StreamRead serves contiguous slices by copying out
+	// of the (possibly wrapped) ring into a fresh buffer.
 	ringMu     sync.Mutex
-	ring       []byte
-	ringStart  int64 // absolute offset of ring[0]
-	ringWrites int64 // cumulative bytes written
+	ring       []byte // length == ringCap once allocated
+	ringPos    int    // next byte to overwrite
+	ringFull   bool   // true once ringWrites ≥ ringCap
+	ringWrites int64  // cumulative bytes written
 
 	// portsMu guards ports. Best-effort port detection: regex on stream.
 	portsMu sync.Mutex
 	ports   []PortSighting
 
+	// Idle-detection state. idleState carries the classifier's current
+	// opinion (StateIdle / StateWorking / …). lastTitleNS is the wall
+	// time of the most recent OSC title change — separate from
+	// lastWriteNS so the osc_title_* strategies can ignore plain output
+	// churn. idleDetection is the compiled per-preset config, resolved
+	// once at spawn and immutable thereafter.
+	idleState     atomic.Pointer[IdleState]
+	idleReason    atomic.Pointer[string]
+	titleMu       sync.RWMutex
+	title         string
+	lastTitleNS   atomic.Int64
+	idleDetection *resolvedIdleDetection
+
 	cleanupMu    sync.Mutex
 	cleanupPaths []string
 	restarting   atomic.Bool
@@ -127,10 +150,36 @@ type Child struct {
 	// exits and calls Start to bring the entry back up. Cleared when the
 	// user explicitly kills the process from the palette.
 	autoRestart atomic.Bool
+
+	// persistFn is set by Session after Spawn registers the entry. The
+	// callback mirrors mutable bits (name, auto-restart) into the
+	// persist store so a restarted patterm can rebuild this entry. Nil
+	// when no persist store is attached (unit tests / non-command
+	// entries).
+	persistMu sync.Mutex
+	persistFn func(*Child)
 }
 
-func (c *Child) SetAutoRestart(v bool) { c.autoRestart.Store(v) }
-func (c *Child) AutoRestart() bool     { return c.autoRestart.Load() }
+func (c *Child) SetAutoRestart(v bool) {
+	c.autoRestart.Store(v)
+	c.firePersist()
+}
+func (c *Child) AutoRestart() bool { return c.autoRestart.Load() }
+
+func (c *Child) setPersistFn(fn func(*Child)) {
+	c.persistMu.Lock()
+	c.persistFn = fn
+	c.persistMu.Unlock()
+}
+
+func (c *Child) firePersist() {
+	c.persistMu.Lock()
+	fn := c.persistFn
+	c.persistMu.Unlock()
+	if fn != nil {
+		fn(c)
+	}
+}
 
 // PortSighting is one entry returned by get_process_ports.
 type PortSighting struct {
@@ -152,7 +201,7 @@ func newChildEntry(id, name string, kind ChildKind, argv, env []string, parentID
 		Kind:      kind,
 		ParentID:  parentID,
 		PresetRef: presetRef,
-		ring:      make([]byte, 0, ringCap),
+		ring:      make([]byte, ringCap),
 	}
 	st := StatusStopped
 	c.status.Store(&st)
@@ -179,7 +228,7 @@ func (c *Child) startPTY(cols, rows uint16) (uint64, error) {
 	}
 	starting := StatusStarting
 	c.status.Store(&starting)
-	p, err := pkgpty.Start(c.Argv, c.Env, cols, rows)
+	p, err := pkgpty.Start(c.Argv, c.Env, c.WorkDir, cols, rows)
 	if err != nil {
 		em.Close()
 		errored := StatusErrored
@@ -254,6 +303,7 @@ func (c *Child) SetName(name string) {
 	c.nameMu.Lock()
 	c.Name = name
 	c.nameMu.Unlock()
+	c.firePersist()
 }
 
 // ScreenVersion returns the current emulator snapshot version, bumped
@@ -298,17 +348,95 @@ func (c *Child) IdleMS() int64 {
 	return (time.Now().UnixNano() - last) / int64(time.Millisecond)
 }
 
+// TitleIdleMS returns how many milliseconds since the OSC window title
+// last changed. 0 means "no title set yet".
+func (c *Child) TitleIdleMS() int64 {
+	last := c.lastTitleNS.Load()
+	if last == 0 {
+		return 0
+	}
+	return (time.Now().UnixNano() - last) / int64(time.Millisecond)
+}
+
+// Title returns the most recent OSC 0/2 title.
+func (c *Child) Title() string {
+	c.titleMu.RLock()
+	defer c.titleMu.RUnlock()
+	return c.title
+}
+
+// recordTitle updates the cached title and bumps lastTitleNS when it
+// actually changes. Called from Session.pumpChild after each PTY chunk
+// — cheap because most chunks don't carry an OSC sequence.
+func (c *Child) recordTitle(newTitle string) {
+	c.titleMu.Lock()
+	if c.title == newTitle {
+		c.titleMu.Unlock()
+		return
+	}
+	c.title = newTitle
+	c.titleMu.Unlock()
+	c.lastTitleNS.Store(time.Now().UnixNano())
+}
+
+// IdleState returns the classifier's current opinion. Empty string
+// (StateUnknown) means the classifier hasn't run yet for this child.
+func (c *Child) IdleState() IdleState {
+	p := c.idleState.Load()
+	if p == nil {
+		return StateUnknown
+	}
+	return *p
+}
+
+// IdleReason returns the human-readable reason the classifier last
+// recorded. Empty when no classification has happened yet.
+func (c *Child) IdleReason() string {
+	p := c.idleReason.Load()
+	if p == nil {
+		return ""
+	}
+	return *p
+}
+
+// setIdleState updates idleState + idleReason. Returns true when the
+// state actually changed (so callers can fan out a notification).
+func (c *Child) setIdleState(s IdleState, reason string) bool {
+	prev := c.IdleState()
+	if prev == s {
+		return false
+	}
+	c.idleState.Store(&s)
+	c.idleReason.Store(&reason)
+	return true
+}
+
+// setIdleDetection installs the resolved per-preset idle-detection
+// config. Called once at spawn; not safe to swap at runtime.
+func (c *Child) setIdleDetection(r *resolvedIdleDetection) {
+	c.idleDetection = r
+}
+
 func (c *Child) recordWrite(chunk []byte) {
 	c.lastWriteNS.Store(time.Now().UnixNano())
 	c.screenVersion.Add(1)
 	c.ringMu.Lock()
-	c.ring = append(c.ring, chunk...)
-	c.ringWrites += int64(len(chunk))
-	if len(c.ring) > ringCap {
-		drop := len(c.ring) - ringCap
-		c.ring = c.ring[drop:]
-		c.ringStart += int64(drop)
+	// Chunks larger than ringCap are tail-truncated — only the last
+	// ringCap bytes of the chunk can survive.
+	src := chunk
+	if len(src) > ringCap {
+		src = src[len(src)-ringCap:]
 	}
+	for written := 0; written < len(src); {
+		n := copy(c.ring[c.ringPos:], src[written:])
+		c.ringPos += n
+		if c.ringPos >= ringCap {
+			c.ringPos = 0
+			c.ringFull = true
+		}
+		written += n
+	}
+	c.ringWrites += int64(len(chunk))
 	c.ringMu.Unlock()
 	c.scanPortsFromChunk(chunk)
 }
@@ -316,6 +444,11 @@ func (c *Child) recordWrite(chunk []byte) {
 // scanPortsFromChunk does best-effort port detection on a PTY chunk.
 // SPEC §7 get_process_ports — no probing, just stream scanning.
 func (c *Child) scanPortsFromChunk(chunk []byte) {
+	// Cheap prefix check: most chunks don't contain a URL. Bail before
+	// running the regex DFA over the whole chunk.
+	if !bytes.Contains(chunk, []byte("http")) {
+		return
+	}
 	matches := portRegex.FindAllSubmatch(chunk, -1)
 	if len(matches) == 0 {
 		return
@@ -364,16 +497,38 @@ func (c *Child) Ports() []PortSighting {
 func (c *Child) StreamRead(since int64) ([]byte, int64) {
 	c.ringMu.Lock()
 	defer c.ringMu.Unlock()
-	if since < c.ringStart {
-		since = c.ringStart
+	end := c.ringWrites
+	var ringStart int64
+	if c.ringFull {
+		ringStart = end - int64(ringCap)
+	}
+	if since < ringStart {
+		since = ringStart
 	}
-	end := c.ringStart + int64(len(c.ring))
 	if since >= end {
 		return nil, end
 	}
-	start := int(since - c.ringStart)
-	out := make([]byte, end-since)
-	copy(out, c.ring[start:])
+	n := int(end - since)
+	out := make([]byte, n)
+	// Locate `since` in the ring. When the buffer hasn't wrapped yet,
+	// bytes 0..ringPos hold writes 0..ringPos. After wrap, ringPos
+	// points at the oldest byte, and the freshest byte is at
+	// (ringPos - 1) mod ringCap.
+	var pos int
+	if c.ringFull {
+		skip := int(since - ringStart) // bytes after the oldest
+		pos = (c.ringPos + skip) % ringCap
+	} else {
+		pos = int(since)
+	}
+	first := ringCap - pos
+	if first > n {
+		first = n
+	}
+	copy(out, c.ring[pos:pos+first])
+	if first < n {
+		copy(out[first:], c.ring[:n-first])
+	}
 	return out, end
 }
 
@@ -395,19 +550,17 @@ func (c *Child) signal(sig syscall.Signal) error {
 // NudgeRedraw asks the child to throw away any diff-based render state
 // and emit a full frame on the next tick. Used after a focus switch so
 // ratatui/ink TUIs re-render coherently against the snapshot we just
-// replayed. We toggle the PTY size by one row so the kernel reliably
-// emits SIGWINCH (TIOCSWINSZ skips the signal if the size didn't
-// change), then send SIGWINCH explicitly for TUIs that miss or coalesce
-// the size-toggled signal. The emulator is left alone — it already
-// matches our intended size and the brief mismatch only affects what the
-// child writes during the second redraw.
+// replayed. Sends an explicit SIGWINCH; TIOCSWINSZ with the same size
+// is a no-op in the kernel, so an explicit signal is what most TUIs
+// actually act on anyway. Avoid resize-toggles here — under a drag-
+// resize the kernel still emits intermediate SIGWINCHes against the
+// host PTY and toggling our child's size on top produces inconsistent
+// grid state.
 func (c *Child) NudgeRedraw(cols, rows uint16) {
 	pty := c.PTY()
 	if pty == nil || rows < 2 {
 		return
 	}
-	_ = pty.Resize(cols, rows-1)
-	_ = pty.Resize(cols, rows)
 	_ = c.signal(syscall.SIGWINCH)
 }
 
@@ -477,25 +630,25 @@ func (c *Child) InjectAsOrchestrator(b []byte) error {
 }
 
 // writeInput is the shared PTY write path used by both injection
-// flavours. Each Enter byte (CR or LF) is split onto its own write
-// with a brief delay so TUI agents with paste-detection (claude,
+// flavours. Agent panes split each Enter byte (CR or LF) onto its own
+// write with a brief delay so TUI agents with paste-detection (claude,
 // codex, opencode) don't coalesce a trailing CR into the text that
-// preceded it. Without the split, `pty.Write([]byte("hello\r"))`
-// arrives at the agent as one read() and gets treated as multi-line
-// pasted content rather than "key Enter".
+// preceded it. Raw terminals and command panes receive the original
+// byte stream in one write; otherwise a multiline paste pays the agent
+// workaround's delay once per line.
 func (c *Child) writeInput(b []byte) error {
 	pty := c.PTY()
 	if pty == nil {
 		return errors.New("child has no pty")
 	}
-	pieces := splitOnEnter(b)
+	pieces := inputWritePieces(c.Kind, b)
 	if len(pieces) <= 1 {
 		_, err := pty.Write(b)
 		return err
 	}
 	for i, piece := range pieces {
-		if i > 0 {
-			time.Sleep(15 * time.Millisecond)
+		if delay := pieceWriteDelay(i, len(pieces), piece); delay > 0 {
+			time.Sleep(delay)
 		}
 		if _, err := pty.Write(piece); err != nil {
 			return err
@@ -504,6 +657,27 @@ func (c *Child) writeInput(b []byte) error {
 	return nil
 }
 
+func inputWritePieces(kind ChildKind, b []byte) [][]byte {
+	if kind != KindAgent {
+		return [][]byte{b}
+	}
+	return splitOnEnter(b)
+}
+
+func pieceWriteDelay(index, total int, piece []byte) time.Duration {
+	if index == 0 {
+		return 0
+	}
+	if index == total-1 && isLoneEnter(piece) {
+		return agentSubmitSettleDelay
+	}
+	return agentInterPieceDelay
+}
+
+func isLoneEnter(piece []byte) bool {
+	return len(piece) == 1 && (piece[0] == '\r' || piece[0] == '\n')
+}
+
 func mintIdentity() string {
 	var buf [12]byte
 	_, _ = rand.Read(buf[:])

internal/app/child_input_test.go

@@ -0,0 +1,90 @@
+package app
+
+import (
+	"bytes"
+	"testing"
+	"time"
+)
+
+func TestInputWritePiecesOnlySplitAgentEnters(t *testing.T) {
+	in := []byte("alpha\nbeta\rgamma")
+	for _, kind := range []ChildKind{KindTerminal, KindCommand} {
+		t.Run(string(kind), func(t *testing.T) {
+			got := inputWritePieces(kind, in)
+			if len(got) != 1 || !bytes.Equal(got[0], in) {
+				t.Fatalf("inputWritePieces(%s) = %#v, want one original chunk", kind, got)
+			}
+		})
+	}
+
+	got := inputWritePieces(KindAgent, in)
+	if len(got) != 5 {
+		t.Fatalf("agent pieces len = %d, want 5 (%#v)", len(got), got)
+	}
+	want := [][]byte{[]byte("alpha"), []byte("\n"), []byte("beta"), []byte("\r"), []byte("gamma")}
+	for i := range want {
+		if !bytes.Equal(got[i], want[i]) {
+			t.Fatalf("agent piece %d = %q, want %q", i, got[i], want[i])
+		}
+	}
+}
+
+func TestPieceWriteDelay(t *testing.T) {
+	cases := []struct {
+		name  string
+		index int
+		total int
+		piece []byte
+		want  time.Duration
+	}{
+		{
+			name:  "first piece",
+			index: 0,
+			total: 3,
+			piece: []byte("body"),
+			want:  0,
+		},
+		{
+			name:  "middle body piece",
+			index: 1,
+			total: 3,
+			piece: []byte("body"),
+			want:  agentInterPieceDelay,
+		},
+		{
+			name:  "final carriage return submit",
+			index: 1,
+			total: 2,
+			piece: []byte("\r"),
+			want:  agentSubmitSettleDelay,
+		},
+		{
+			name:  "final newline submit",
+			index: 1,
+			total: 2,
+			piece: []byte("\n"),
+			want:  agentSubmitSettleDelay,
+		},
+		{
+			name:  "final non-enter piece",
+			index: 2,
+			total: 3,
+			piece: []byte("tail"),
+			want:  agentInterPieceDelay,
+		},
+		{
+			name:  "standalone enter fast path",
+			index: 0,
+			total: 1,
+			piece: []byte("\r"),
+			want:  0,
+		},
+	}
+	for _, tc := range cases {
+		t.Run(tc.name, func(t *testing.T) {
+			if got := pieceWriteDelay(tc.index, tc.total, tc.piece); got != tc.want {
+				t.Fatalf("pieceWriteDelay(%d, %d, %q) = %s, want %s", tc.index, tc.total, tc.piece, got, tc.want)
+			}
+		})
+	}
+}

internal/app/chrome_model.go

@@ -0,0 +1,80 @@
+package app
+
+import "github.com/hjbdev/patterm/internal/scratchpad"
+
+// chromeModel is the semantic host chrome state. Renderers continue to own
+// ANSI output; this model is the serializable shape a client can draw locally.
+type chromeModel struct {
+	ProjectKey    string            `json:"project_key"`
+	ProjectName   string            `json:"project_name,omitempty"`
+	FocusedID     string            `json:"focused_id,omitempty"`
+	FocusedPad    string            `json:"focused_pad,omitempty"`
+	ActiveAgentID string            `json:"active_agent_id,omitempty"`
+	Tabs          []childModel      `json:"tabs"`
+	Processes     []childModel      `json:"processes"`
+	AgentTree     []childModel      `json:"agent_tree"`
+	Sidebar       []navEntryModel   `json:"sidebar"`
+	Scratchpads   []scratchpadModel `json:"scratchpads"`
+}
+
+type childModel struct {
+	ID       string `json:"id"`
+	Name     string `json:"name"`
+	Kind     string `json:"kind"`
+	ParentID string `json:"parent_id,omitempty"`
+	Status   string `json:"status"`
+	Owner    string `json:"owner"`
+}
+
+type navEntryModel struct {
+	ChildID string `json:"child_id,omitempty"`
+	Pad     string `json:"pad,omitempty"`
+}
+
+type scratchpadModel struct {
+	Name string `json:"name"`
+}
+
+func buildChromeModel(projectKey string, view ClientView, children []*Child, pads []scratchpad.Entry) chromeModel {
+	active := view.ActiveAgentID
+	if active == "" {
+		active = activeRootID(children, view.FocusedID)
+	}
+	model := chromeModel{
+		ProjectKey:    projectKey,
+		ProjectName:   view.ProjectName,
+		FocusedID:     view.FocusedID,
+		FocusedPad:    view.FocusedPad,
+		ActiveAgentID: active,
+	}
+	for _, c := range runningTopLevels(children) {
+		model.Tabs = append(model.Tabs, serializeChildModel(c))
+	}
+	for _, c := range processList(children) {
+		model.Processes = append(model.Processes, serializeChildModel(c))
+	}
+	for _, c := range visibleAgentTree(children, active) {
+		model.AgentTree = append(model.AgentTree, serializeChildModel(c))
+	}
+	for _, n := range sidebarNav(children, active, pads) {
+		model.Sidebar = append(model.Sidebar, navEntryModel{ChildID: n.childID, Pad: n.pad})
+	}
+	for _, p := range pads {
+		model.Scratchpads = append(model.Scratchpads, scratchpadModel{Name: p.Name})
+	}
+	return model
+}
+
+func serializeChildModel(c *Child) childModel {
+	if c == nil {
+		return childModel{}
+	}
+	return childModel{
+		ID:       c.ID,
+		Name:     c.DisplayName(),
+		Kind:     string(c.Kind),
+		ParentID: c.ParentID,
+		Status:   string(c.Status()),
+		Owner:    string(c.Owner()),
+	}
+}

internal/app/chrome_model_test.go

@@ -0,0 +1,24 @@
+package app
+
+import "testing"
+
+func TestBuildChromeModelSeparatesProcessesTabsAndSidebar(t *testing.T) {
+	running := StatusRunning
+	proc := testProcess("p1", "server", running)
+	agent := testAgent("a1", "codex", "", running)
+	sub := testAgent("a2", "worker", "a1", running)
+
+	model := buildChromeModel("project", ClientView{FocusedID: "p1", ActiveAgentID: "a1"}, []*Child{proc, agent, sub}, nil)
+	if len(model.Tabs) != 1 || model.Tabs[0].ID != "a1" {
+		t.Fatalf("tabs = %#v, want only top-level agent", model.Tabs)
+	}
+	if len(model.Processes) != 1 || model.Processes[0].ID != "p1" {
+		t.Fatalf("processes = %#v, want process section", model.Processes)
+	}
+	if len(model.AgentTree) != 2 || model.AgentTree[0].ID != "a1" || model.AgentTree[1].ID != "a2" {
+		t.Fatalf("agent tree = %#v", model.AgentTree)
+	}
+	if len(model.Sidebar) != 3 || model.Sidebar[0].ChildID != "p1" || model.Sidebar[1].ChildID != "a1" {
+		t.Fatalf("sidebar = %#v", model.Sidebar)
+	}
+}

internal/app/classifier.go

@@ -0,0 +1,102 @@
+package app
+
+import (
+	"context"
+	"time"
+)
+
+// classifierTickInterval is how often the per-session classifier wakes
+// up to re-evaluate every child's state. 250ms is fast enough that
+// the sidebar badge looks live, slow enough that the cost is invisible
+// even with dozens of children.
+const classifierTickInterval = 250 * time.Millisecond
+
+// classifierTailBytes is the size of the ring-buffer tail the
+// classifier scans for promoter regexes. Big enough to catch a multi-
+// line "Approve?" prompt, small enough that we don't pay for a full
+// 1 MiB regex scan every tick.
+const classifierTailBytes = 4096
+
+// runClassifier loops over every live child every classifierTickInterval
+// and updates IdleState when it changes. It runs until ctx is cancelled
+// (the host shutdown path cancels). One goroutine per Session is plenty
+// — the work is cheap (atomic loads + ~4 KiB regex scan per child).
+func (s *Session) runClassifier(ctx context.Context) {
+	ticker := time.NewTicker(classifierTickInterval)
+	defer ticker.Stop()
+	for {
+		select {
+		case <-ctx.Done():
+			return
+		case <-ticker.C:
+			s.classifyAll()
+		}
+	}
+}
+
+func (s *Session) classifyAll() {
+	for _, c := range s.Children() {
+		s.classifyOne(c)
+	}
+}
+
+func (s *Session) classifyOne(c *Child) {
+	st := c.Status()
+	exited := st == StatusExited || st == StatusErrored
+	exitNonZero := false
+	if exited {
+		exitNonZero = c.ExitCode() != 0
+	}
+	idleMS := c.IdleMS()
+	titleIdleMS := c.TitleIdleMS()
+	title := c.Title()
+	tail := stripANSIBytes(nil, c.tailBytes(classifierTailBytes))
+	var screen []byte
+	if em := c.Emulator(); em != nil {
+		if txt, err := em.ScreenText(); err == nil {
+			screen = []byte(txt)
+		}
+	}
+	state, reason := classify(c.idleDetection, exited, exitNonZero, idleMS, titleIdleMS, title, tail, screen)
+	if c.setIdleState(state, reason) {
+		s.emitStateChanged(c.ID, state)
+	}
+}
+
+// tailBytes returns up to n bytes from the end of the ring buffer.
+// Safe to call from the classifier goroutine while pumpChild writes
+// from another goroutine — both serialise on ringMu.
+func (c *Child) tailBytes(n int) []byte {
+	c.ringMu.Lock()
+	defer c.ringMu.Unlock()
+	have := int64(ringCap)
+	if !c.ringFull {
+		have = c.ringWrites
+	}
+	if have == 0 {
+		return nil
+	}
+	want := int64(n)
+	if want > have {
+		want = have
+	}
+	out := make([]byte, want)
+	// The ring layout matches StreamRead: when not full, byte k lives
+	// at index k; when full, the oldest byte sits at ringPos and the
+	// newest at (ringPos-1) mod ringCap.
+	if !c.ringFull {
+		copy(out, c.ring[c.ringWrites-want:c.ringWrites])
+		return out
+	}
+	// Tail starts `want` bytes back from the write head.
+	start := (c.ringPos - int(want) + ringCap) % ringCap
+	first := ringCap - start
+	if first > int(want) {
+		first = int(want)
+	}
+	copy(out, c.ring[start:start+first])
+	if first < int(want) {
+		copy(out[first:], c.ring[:int(want)-first])
+	}
+	return out
+}

internal/app/client_net.go

@@ -0,0 +1,677 @@
+package app
+
+import (
+	"context"
+	"encoding/json"
+	"errors"
+	"fmt"
+	"io"
+	"net"
+	"os"
+	"os/exec"
+	"os/signal"
+	"strings"
+	"sync"
+	"syscall"
+	"time"
+
+	cpty "github.com/creack/pty"
+	"golang.org/x/term"
+
+	"github.com/hjbdev/patterm/internal/protocol"
+)
+
+const (
+	clientKeyCtrlK       byte = 0x0b
+	clientKeyCtrlBracket byte = 0x1d
+)
+
+type ClientOptions struct {
+	ProjectDir string
+	Transport  protocol.Transport
+	Stdin      io.Reader
+	Stdout     io.Writer
+	RawMode    bool
+	AutoStart  bool
+	Token      string
+	Cols       uint16
+	Rows       uint16
+}
+
+func RunAttachedClient(ctx context.Context, opts ClientOptions) error {
+	if opts.ProjectDir == "" {
+		cwd, err := os.Getwd()
+		if err != nil {
+			return err
+		}
+		opts.ProjectDir = cwd
+	}
+	if opts.Stdin == nil {
+		opts.Stdin = os.Stdin
+	}
+	if opts.Stdout == nil {
+		opts.Stdout = os.Stdout
+	}
+	if opts.Transport == nil {
+		t, err := dialDaemonTransport(opts.ProjectDir, opts.AutoStart)
+		if err != nil {
+			return err
+		}
+		opts.Transport = t
+		defer t.Close()
+	}
+	if opts.Cols == 0 || opts.Rows == 0 {
+		opts.Cols, opts.Rows = clientHostSize(opts.Stdin)
+	}
+	c := newNetClient(opts)
+	return c.run(ctx)
+}
+
+func DialTCPTransport(addr string) (protocol.Transport, error) {
+	conn, err := net.Dial("tcp", addr)
+	if err != nil {
+		return nil, err
+	}
+	return protocol.NewConnTransport(conn), nil
+}
+
+func dialDaemonTransport(projectDir string, autoStart bool) (protocol.Transport, error) {
+	socket, _, err := RuntimeDaemonPaths()
+	if err != nil {
+		return nil, err
+	}
+	conn, err := net.Dial("unix", socket)
+	if err == nil {
+		return protocol.NewConnTransport(conn), nil
+	}
+	if !autoStart {
+		return nil, err
+	}
+	if err := startDaemonProcess(projectDir); err != nil {
+		return nil, err
+	}
+	deadline := time.Now().Add(5 * time.Second)
+	var last error
+	for time.Now().Before(deadline) {
+		conn, err = net.Dial("unix", socket)
+		if err == nil {
+			return protocol.NewConnTransport(conn), nil
+		}
+		last = err
+		time.Sleep(50 * time.Millisecond)
+	}
+	return nil, fmt.Errorf("daemon did not become ready: %w", last)
+}
+
+func startDaemonProcess(projectDir string) error {
+	exe, err := os.Executable()
+	if err != nil {
+		return err
+	}
+	cmd := exec.Command(exe, "daemon", "--project", projectDir)
+	devNull, err := os.OpenFile(os.DevNull, os.O_RDWR, 0)
+	if err == nil {
+		defer devNull.Close()
+		cmd.Stdin = devNull
+		cmd.Stdout = devNull
+		cmd.Stderr = devNull
+	}
+	cmd.Env = os.Environ()
+	if err := cmd.Start(); err != nil {
+		return err
+	}
+	return cmd.Process.Release()
+}
+
+type netClient struct {
+	t          protocol.Transport
+	in         io.Reader
+	out        io.Writer
+	raw        bool
+	projectDir string
+	token      string
+	layout     terminalLayout
+
+	mu        sync.Mutex
+	focusedID string
+	paneSize  protocol.Size
+	ownerView bool
+	chrome    chromeModel
+	renderer  *viewportRenderer
+	palette   *clientCommandPrompt
+}
+
+type clientCommandPrompt struct {
+	buf []byte
+}
+
+func newNetClient(opts ClientOptions) *netClient {
+	layout := newTerminalLayout(opts.Cols, opts.Rows)
+	return &netClient{
+		t:          opts.Transport,
+		in:         opts.Stdin,
+		out:        opts.Stdout,
+		raw:        opts.RawMode,
+		projectDir: opts.ProjectDir,
+		token:      opts.Token,
+		layout:     layout,
+		renderer:   newViewportRenderer(layout),
+	}
+}
+
+func (c *netClient) run(ctx context.Context) error {
+	ctx, cancel := context.WithCancel(ctx)
+	defer cancel()
+	var restore *term.State
+	if c.raw {
+		if f, ok := c.in.(*os.File); ok && term.IsTerminal(int(f.Fd())) {
+			st, err := term.MakeRaw(int(f.Fd()))
+			if err != nil {
+				return err
+			}
+			restore = st
+			defer term.Restore(int(f.Fd()), restore)
+		}
+	}
+	c.enterScreen()
+	defer c.leaveScreen()
+
+	if err := c.sendAttach(); err != nil {
+		return err
+	}
+	errCh := make(chan error, 2)
+	go func() { errCh <- c.recvLoop(ctx, cancel) }()
+	go func() { errCh <- c.stdinLoop(ctx, cancel) }()
+	if f, ok := c.in.(*os.File); ok && term.IsTerminal(int(f.Fd())) {
+		winch := make(chan os.Signal, 1)
+		signal.Notify(winch, syscall.SIGWINCH)
+		defer signal.Stop(winch)
+		go func() {
+			for {
+				select {
+				case <-ctx.Done():
+					return
+				case <-winch:
+					cols, rows := clientHostSize(c.in)
+					_ = c.resize(cols, rows)
+					c.enterScreen()
+					c.drawChrome()
+				}
+			}
+		}()
+	}
+	select {
+	case <-ctx.Done():
+		_ = c.t.Close()
+		return nil
+	case err := <-errCh:
+		cancel()
+		_ = c.t.Close()
+		if errors.Is(err, io.EOF) || errors.Is(err, protocol.ErrTransportClosed) {
+			return nil
+		}
+		return err
+	}
+}
+
+func (c *netClient) sendAttach() error {
+	f, err := protocol.NewFrame(protocol.FrameAttach, protocol.Attach{
+		ProjectPath: c.projectPath(),
+		Token:       c.token,
+		TermSize: protocol.Size{
+			Cols: c.layout.childCols(),
+			Rows: c.layout.childRows(),
+		},
+	})
+	if err != nil {
+		return err
+	}
+	return c.t.Send(f)
+}
+
+func (c *netClient) projectPath() string {
+	return c.projectDir
+}
+
+func (c *netClient) recvLoop(ctx context.Context, cancel func()) error {
+	for {
+		select {
+		case <-ctx.Done():
+			return nil
+		default:
+		}
+		f, err := c.t.Recv()
+		if err != nil {
+			return err
+		}
+		if err := c.handleFrame(f); err != nil {
+			return err
+		}
+		if f.Type == protocol.FrameDetach {
+			cancel()
+			return nil
+		}
+	}
+}
+
+func (c *netClient) handleFrame(f protocol.Frame) error {
+	switch f.Type {
+	case protocol.FrameError:
+		msg, _ := protocol.Decode[protocol.Error](f)
+		if msg.Message == "" {
+			msg.Message = "daemon error"
+		}
+		return fmt.Errorf("%s", msg.Message)
+	case protocol.FrameHello:
+		return nil
+	case protocol.FrameProjectList:
+		return nil
+	case protocol.FrameChrome:
+		msg, err := protocol.Decode[protocol.Chrome](f)
+		if err != nil {
+			return err
+		}
+		var model chromeModel
+		if err := json.Unmarshal(msg.Model, &model); err != nil {
+			return err
+		}
+		c.mu.Lock()
+		c.chrome = model
+		if model.FocusedID != "" {
+			c.focusedID = model.FocusedID
+		}
+		c.mu.Unlock()
+		c.drawChrome()
+	case protocol.FramePaneSnapshot:
+		msg, err := protocol.Decode[protocol.PaneSnapshot](f)
+		if err != nil {
+			return err
+		}
+		c.mu.Lock()
+		c.focusedID = msg.PaneID
+		c.paneSize = msg.Size
+		c.ownerView = msg.DisplayOwner
+		c.renderer = newViewportRenderer(c.renderLayoutLocked(msg.Size))
+		renderer := c.renderer
+		c.mu.Unlock()
+		c.clearViewport()
+		c.drawChrome()
+		c.writeWrapped(renderer.Render(msg.Bytes))
+	case protocol.FramePaneChunk:
+		msg, err := protocol.Decode[protocol.PaneChunk](f)
+		if err != nil {
+			return err
+		}
+		c.mu.Lock()
+		focused := c.focusedID
+		renderer := c.renderer
+		c.paneSize = msg.Size
+		c.ownerView = msg.DisplayOwner
+		if renderer != nil && (msg.Size.Cols != 0 || msg.Size.Rows != 0) {
+			renderer.SetLayout(c.renderLayoutLocked(msg.Size))
+		}
+		c.mu.Unlock()
+		if msg.PaneID == focused && renderer != nil {
+			c.writeWrapped(renderer.Render(msg.Bytes))
+		}
+	case protocol.FrameLifecycle:
+		// The daemon follows lifecycle changes with chrome/snapshot updates
+		// when focus changes. Keep this as a wake point for future richer
+		// client-side state without blocking the frame stream.
+		return nil
+	}
+	return nil
+}
+
+func (c *netClient) stdinLoop(ctx context.Context, cancel func()) error {
+	buf := make([]byte, 4096)
+	for {
+		n, err := c.in.Read(buf)
+		if n > 0 {
+			if done, perr := c.processInput(buf[:n]); perr != nil || done {
+				cancel()
+				return perr
+			}
+		}
+		if err != nil {
+			if errors.Is(err, io.EOF) {
+				return nil
+			}
+			return err
+		}
+		select {
+		case <-ctx.Done():
+			return nil
+		default:
+		}
+	}
+}
+
+func (c *netClient) processInput(chunk []byte) (bool, error) {
+	c.mu.Lock()
+	if c.palette != nil {
+		p := c.palette
+		c.mu.Unlock()
+		return c.processPaletteInput(p, chunk)
+	}
+	c.mu.Unlock()
+
+	forward := make([]byte, 0, len(chunk))
+	flush := func() error {
+		if len(forward) == 0 {
+			return nil
+		}
+		c.mu.Lock()
+		paneID := c.focusedID
+		c.mu.Unlock()
+		if paneID != "" {
+			f, err := protocol.NewFrame(protocol.FrameInput, protocol.Input{PaneID: paneID, Bytes: append([]byte(nil), forward...)})
+			if err != nil {
+				return err
+			}
+			if err := c.t.Send(f); err != nil {
+				return err
+			}
+		}
+		forward = forward[:0]
+		return nil
+	}
+	for _, b := range chunk {
+		switch b {
+		case clientKeyCtrlBracket:
+			if err := flush(); err != nil {
+				return false, err
+			}
+			return true, c.sendDetach()
+		case clientKeyCtrlK:
+			if err := flush(); err != nil {
+				return false, err
+			}
+			c.mu.Lock()
+			c.palette = &clientCommandPrompt{}
+			c.mu.Unlock()
+			c.drawPrompt()
+		case 0x17: // Ctrl-W: previous focus
+			if err := flush(); err != nil {
+				return false, err
+			}
+			_ = c.focusRelative(-1)
+		case 0x13: // Ctrl-S: next focus
+			if err := flush(); err != nil {
+				return false, err
+			}
+			_ = c.focusRelative(1)
+		default:
+			forward = append(forward, b)
+		}
+	}
+	return false, flush()
+}
+
+func (c *netClient) processPaletteInput(p *clientCommandPrompt, chunk []byte) (bool, error) {
+	for _, b := range chunk {
+		switch b {
+		case 0x1b: // ESC
+			c.mu.Lock()
+			c.palette = nil
+			c.mu.Unlock()
+			c.drawChrome()
+			return false, nil
+		case 'd':
+			if len(p.buf) == 0 {
+				c.mu.Lock()
+				c.palette = nil
+				c.mu.Unlock()
+				return true, c.sendDetach()
+			}
+			p.buf = append(p.buf, b)
+		case '\r', '\n':
+			command := strings.TrimSpace(string(p.buf))
+			c.mu.Lock()
+			c.palette = nil
+			c.mu.Unlock()
+			if command == "" {
+				c.drawChrome()
+				return false, nil
+			}
+			return false, c.sendSpawnCommand(command)
+		case 0x7f, 0x08:
+			if len(p.buf) > 0 {
+				p.buf = p.buf[:len(p.buf)-1]
+			}
+			c.drawPrompt()
+		default:
+			if b >= 0x20 {
+				p.buf = append(p.buf, b)
+				c.drawPrompt()
+			}
+		}
+	}
+	return false, nil
+}
+
+func (c *netClient) sendDetach() error {
+	f, err := protocol.NewFrame(protocol.FrameDetach, protocol.Detach{})
+	if err != nil {
+		return err
+	}
+	return c.t.Send(f)
+}
+
+func (c *netClient) sendSpawnCommand(command string) error {
+	data, err := json.Marshal(map[string]any{
+		"argv":  []string{command},
+		"name":  command,
+		"shell": true,
+	})
+	if err != nil {
+		return err
+	}
+	f, err := protocol.NewFrame(protocol.FramePaletteCommand, protocol.PaletteCommand{
+		Kind: "spawn_command",
+		Data: data,
+	})
+	if err != nil {
+		return err
+	}
+	return c.t.Send(f)
+}
+
+func (c *netClient) focusRelative(delta int) error {
+	c.mu.Lock()
+	model := c.chrome
+	current := c.focusedID
+	c.mu.Unlock()
+	ids := make([]string, 0, len(model.Processes)+len(model.AgentTree)+len(model.Tabs))
+	for _, n := range model.Sidebar {
+		if n.ChildID != "" {
+			ids = append(ids, n.ChildID)
+		}
+	}
+	if len(ids) == 0 {
+		for _, p := range model.Processes {
+			ids = append(ids, p.ID)
+		}
+		for _, p := range model.Tabs {
+			ids = append(ids, p.ID)
+		}
+	}
+	if len(ids) == 0 {
+		return nil
+	}
+	idx := 0
+	for i, id := range ids {
+		if id == current {
+			idx = i
+			break
+		}
+	}
+	idx = (idx + delta + len(ids)) % len(ids)
+	f, err := protocol.NewFrame(protocol.FrameFocus, protocol.Focus{PaneID: ids[idx]})
+	if err != nil {
+		return err
+	}
+	return c.t.Send(f)
+}
+
+func (c *netClient) resize(cols, rows uint16) error {
+	c.mu.Lock()
+	c.layout = newTerminalLayout(cols, rows)
+	if c.renderer != nil {
+		c.renderer.SetLayout(c.renderLayoutLocked(c.paneSize))
+	}
+	size := protocol.Size{Cols: c.layout.childCols(), Rows: c.layout.childRows()}
+	c.mu.Unlock()
+	f, err := protocol.NewFrame(protocol.FrameResize, protocol.Resize{Size: size})
+	if err != nil {
+		return err
+	}
+	return c.t.Send(f)
+}
+
+func (c *netClient) renderLayoutLocked(size protocol.Size) terminalLayout {
+	l := c.layout
+	if size.Cols != 0 && size.Cols < l.mainCols {
+		l.mainCols = size.Cols
+	}
+	if size.Rows != 0 && size.Rows < l.mainRows {
+		l.mainRows = size.Rows
+	}
+	return l
+}
+
+func (c *netClient) enterScreen() {
+	_, _ = c.out.Write([]byte("\x1b[?1049h\x1b[H\x1b[2J\x1b[?25h\x1b[?1000h\x1b[?1006h"))
+	c.installScrollRegion()
+}
+
+func (c *netClient) leaveScreen() {
+	_, _ = c.out.Write([]byte("\x1b[r\x1b[?6l\x1b[?1006l\x1b[?1000l\x1b[?25h\x1b[?1049l"))
+}
+
+func (c *netClient) installScrollRegion() {
+	mainBottom := int(c.layout.statusRow) - statusRows
+	if mainBottom < int(c.layout.mainTop) {
+		return
+	}
+	fmt.Fprintf(c.out, "\x1b[?6l\x1b[%d;%dr\x1b[%d;%dH",
+		int(c.layout.mainTop), mainBottom,
+		int(c.layout.mainTop), int(c.layout.mainLeft))
+}
+
+func (c *netClient) clearViewport() {
+	for row := int(c.layout.mainTop); row < int(c.layout.statusRow); row++ {
+		fmt.Fprintf(c.out, "\x1b[%d;%dH\x1b[%dX", row, int(c.layout.mainLeft), int(c.layout.childCols()))
+	}
+	fmt.Fprintf(c.out, "\x1b[%d;%dH", int(c.layout.mainTop), int(c.layout.mainLeft))
+}
+
+func (c *netClient) writeWrapped(out []byte) {
+	if len(out) == 0 {
+		return
+	}
+	wrapped := make([]byte, 0, len(out)+10)
+	wrapped = append(wrapped, "\x1b[?7l"...)
+	wrapped = append(wrapped, out...)
+	wrapped = append(wrapped, "\x1b[?7h"...)
+	_, _ = c.out.Write(wrapped)
+}
+
+func (c *netClient) drawChrome() {
+	c.mu.Lock()
+	model := c.chrome
+	prompt := c.palette
+	c.mu.Unlock()
+	var b strings.Builder
+	width := int(c.layout.childCols())
+	fmt.Fprintf(&b, "\x1b[1;1H\x1b[%dX\x1b[2;1H\x1b[%dX\x1b[3;1H\x1b[%dX", width, width, width)
+	if len(model.Tabs) == 0 {
+		fmt.Fprintf(&b, "\x1b[1;2H%s+ new%s", styleDim, styleReset)
+	} else {
+		col := 1
+		for _, tab := range model.Tabs {
+			label := fitName(tab.Name, 18)
+			style := styleHint
+			if tab.ID == model.ActiveAgentID || tab.ID == model.FocusedID {
+				style = styleActive
+			}
+			fmt.Fprintf(&b, "\x1b[1;%dH%s %s %s", col, style, label, styleReset)
+			col += visibleLen(label) + 3
+			if col >= width {
+				break
+			}
+		}
+	}
+	fmt.Fprintf(&b, "\x1b[3;1H%s%s%s", styleBorder, strings.Repeat("─", width), styleReset)
+	if c.layout.sidebarVisible {
+		c.appendSidebar(&b, model)
+	}
+	status := "Ctrl-K command palette  ·  Ctrl-] detach"
+	if model.FocusedID != "" {
+		status = fmt.Sprintf("%s · %s", model.FocusedID, status)
+	}
+	c.mu.Lock()
+	size := c.paneSize
+	ownerView := c.ownerView
+	c.mu.Unlock()
+	if model.FocusedID != "" && !ownerView && size.Cols != 0 && size.Rows != 0 {
+		status = fmt.Sprintf("viewing at owner size %dx%d · %s", size.Cols, size.Rows, status)
+	}
+	if prompt != nil {
+		status = "command: " + string(prompt.buf)
+	}
+	fmt.Fprintf(&b, "\x1b[%d;1H\x1b[7m%s%s", int(c.layout.statusRow), fitName(status, int(c.layout.hostCols)), styleReset)
+	_, _ = c.out.Write([]byte(b.String()))
+}
+
+func (c *netClient) appendSidebar(b *strings.Builder, model chromeModel) {
+	border := int(c.layout.sidebarLeft) - 1
+	for row := 1; row <= int(c.layout.statusRow)-1; row++ {
+		fmt.Fprintf(b, "\x1b[%d;%dH%s│%s", row, border, styleBorder, styleReset)
+	}
+	col := int(c.layout.sidebarLeft)
+	row := 1
+	write := func(text string) {
+		if row >= int(c.layout.statusRow) {
+			return
+		}
+		fmt.Fprintf(b, "\x1b[%d;%dH%-*s", row, col, int(c.layout.sidebarWidth)-1, fitName(text, int(c.layout.sidebarWidth)-1))
+		row++
+	}
+	write(styleActive + "Processes" + styleReset)
+	for _, p := range model.Processes {
+		prefix := "  "
+		if p.ID == model.FocusedID {
+			prefix = "▎ "
+		}
+		write(prefix + p.Name)
+	}
+	row++
+	write(styleActive + "Agent Tree" + styleReset)
+	for _, p := range model.AgentTree {
+		prefix := "  "
+		if p.ID == model.FocusedID {
+			prefix = "▎ "
+		}
+		write(prefix + p.Name)
+	}
+	row++
+	write(styleActive + "Scratchpads" + styleReset)
+	for _, p := range model.Scratchpads {
+		write("  " + p.Name)
+	}
+}
+
+func (c *netClient) drawPrompt() {
+	c.drawChrome()
+}
+
+func clientHostSize(r io.Reader) (cols, rows uint16) {
+	if f, ok := r.(*os.File); ok {
+		ws, err := cpty.GetsizeFull(f)
+		if err == nil && ws.Cols > 0 && ws.Rows > 0 {
+			return ws.Cols, ws.Rows
+		}
+	}
+	return 120, 40
+}

internal/app/client_net_test.go

@@ -0,0 +1,157 @@
+package app
+
+import (
+	"bytes"
+	"context"
+	"encoding/json"
+	"io"
+	"sync"
+	"testing"
+	"time"
+
+	"github.com/hjbdev/patterm/internal/protocol"
+)
+
+func TestNetClientFrameLoopSendsFocusedInput(t *testing.T) {
+	clientT, daemonT := protocol.NewLoopbackPair()
+	inR, inW := ioPipe(t)
+	out := &lockedBuffer{}
+
+	gotInput := make(chan protocol.Input, 1)
+	errCh := make(chan error, 1)
+	go func() {
+		f, err := daemonT.Recv()
+		if err != nil {
+			errCh <- err
+			return
+		}
+		if f.Type != protocol.FrameAttach {
+			t.Errorf("first frame = %s, want attach", f.Type)
+			errCh <- nil
+			return
+		}
+		sendTestFrame(t, daemonT, protocol.FrameHello, protocol.Hello{Version: 1, ClientID: "test", ProjectKey: "project"})
+		sendTestFrame(t, daemonT, protocol.FrameProjectList, protocol.ProjectList{})
+		model := chromeModel{
+			ProjectKey: "project",
+			FocusedID:  "p1",
+			Processes:  []childModel{{ID: "p1", Name: "shell", Kind: string(KindCommand), Status: string(StatusRunning)}},
+			Sidebar:    []navEntryModel{{ChildID: "p1"}},
+		}
+		sendTestFrame(t, daemonT, protocol.FrameChrome, protocol.Chrome{ProjectKey: "project", Model: mustMarshalTest(t, model)})
+		sendTestFrame(t, daemonT, protocol.FramePaneSnapshot, protocol.PaneSnapshot{PaneID: "p1", Bytes: []byte("READY")})
+		for {
+			f, err := daemonT.Recv()
+			if err != nil {
+				errCh <- err
+				return
+			}
+			if f.Type != protocol.FrameInput {
+				continue
+			}
+			input, err := protocol.Decode[protocol.Input](f)
+			if err != nil {
+				errCh <- err
+				return
+			}
+			gotInput <- input
+			_ = daemonT.Close()
+			errCh <- nil
+			return
+		}
+	}()
+
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+	runCh := make(chan error, 1)
+	go func() {
+		runCh <- RunAttachedClient(ctx, ClientOptions{
+			Transport: clientT,
+			Stdin:     inR,
+			Stdout:    out,
+			Cols:      80,
+			Rows:      24,
+		})
+	}()
+
+	deadline := time.Now().Add(3 * time.Second)
+	for time.Now().Before(deadline) && !bytes.Contains(out.Bytes(), []byte("READY")) {
+		time.Sleep(10 * time.Millisecond)
+	}
+	if !bytes.Contains(out.Bytes(), []byte("READY")) {
+		t.Fatalf("snapshot was not rendered before input; output=%q", out.String())
+	}
+	if _, err := inW.Write([]byte("echo hi\r")); err != nil {
+		t.Fatalf("write stdin: %v", err)
+	}
+	select {
+	case input := <-gotInput:
+		if input.PaneID != "p1" || string(input.Bytes) != "echo hi\r" {
+			t.Fatalf("input = %#v", input)
+		}
+	case <-time.After(3 * time.Second):
+		t.Fatalf("client did not forward input")
+	}
+	cancel()
+	_ = inW.Close()
+	select {
+	case err := <-runCh:
+		if err != nil {
+			t.Fatalf("client run: %v", err)
+		}
+	case <-time.After(3 * time.Second):
+		t.Fatalf("client did not stop")
+	}
+	if err := <-errCh; err != nil && err != protocol.ErrTransportClosed {
+		t.Fatalf("daemon side: %v", err)
+	}
+}
+
+type lockedBuffer struct {
+	mu sync.Mutex
+	b  bytes.Buffer
+}
+
+func (b *lockedBuffer) Write(p []byte) (int, error) {
+	b.mu.Lock()
+	defer b.mu.Unlock()
+	return b.b.Write(p)
+}
+
+func (b *lockedBuffer) Bytes() []byte {
+	b.mu.Lock()
+	defer b.mu.Unlock()
+	return append([]byte(nil), b.b.Bytes()...)
+}
+
+func (b *lockedBuffer) String() string {
+	b.mu.Lock()
+	defer b.mu.Unlock()
+	return b.b.String()
+}
+
+func ioPipe(t *testing.T) (*io.PipeReader, *io.PipeWriter) {
+	t.Helper()
+	r, w := io.Pipe()
+	return r, w
+}
+
+func sendTestFrame[T any](t *testing.T, tr protocol.Transport, typ protocol.FrameType, payload T) {
+	t.Helper()
+	f, err := protocol.NewFrame(typ, payload)
+	if err != nil {
+		t.Fatalf("frame %s: %v", typ, err)
+	}
+	if err := tr.Send(f); err != nil {
+		t.Fatalf("send %s: %v", typ, err)
+	}
+}
+
+func mustMarshalTest(t *testing.T, v any) []byte {
+	t.Helper()
+	b, err := json.Marshal(v)
+	if err != nil {
+		t.Fatalf("marshal: %v", err)
+	}
+	return b
+}

internal/app/client_subscriber.go

@@ -0,0 +1,135 @@
+package app
+
+import (
+	"encoding/json"
+	"sync"
+
+	"github.com/hjbdev/patterm/internal/protocol"
+)
+
+const defaultClientSubscriberQueue = 256
+
+// clientSubscriber is the daemon-to-client event bridge. Unlike daemon-local
+// listeners such as timers, debug capture, and waiters, it never blocks the PTY
+// pump: PTY chunks are copied before enqueue, and overflow marks the pane as
+// needing a fresh snapshot.
+type clientSubscriber struct {
+	projectKey string
+	project    *Project
+	clientID   string
+	frames     chan protocol.Frame
+
+	mu               sync.Mutex
+	snapshotRequired map[string]bool
+	lifecycleDirty   bool
+}
+
+func newClientSubscriber(project *Project, clientID string, size int) *clientSubscriber {
+	if size <= 0 {
+		size = defaultClientSubscriberQueue
+	}
+	projectKey := ""
+	if project != nil {
+		projectKey = project.Key
+	}
+	return &clientSubscriber{
+		projectKey:       projectKey,
+		project:          project,
+		clientID:         clientID,
+		frames:           make(chan protocol.Frame, size),
+		snapshotRequired: make(map[string]bool),
+		lifecycleDirty:   false,
+	}
+}
+
+func (s *clientSubscriber) Recv() (protocol.Frame, bool) {
+	f, ok := <-s.frames
+	return f, ok
+}
+
+func (s *clientSubscriber) SnapshotRequired(childID string) bool {
+	s.mu.Lock()
+	defer s.mu.Unlock()
+	return s.snapshotRequired[childID]
+}
+
+func (s *clientSubscriber) OnChildSpawned(c *Child) {
+	s.sendLifecycle(protocol.LifecycleSpawned, c, "")
+}
+
+func (s *clientSubscriber) OnChildExited(c *Child) {
+	s.sendLifecycle(protocol.LifecycleExited, c, "")
+}
+
+func (s *clientSubscriber) OnChildClosed(id string) {
+	s.sendFrame(protocol.Frame{Type: protocol.FrameLifecycle, Payload: mustJSON(protocol.Lifecycle{
+		Kind:       protocol.LifecycleClosed,
+		ProjectKey: s.projectKey,
+		ChildID:    id,
+	})})
+}
+
+func (s *clientSubscriber) OnChildStateChanged(id string, state IdleState) {
+	s.sendFrame(protocol.Frame{Type: protocol.FrameLifecycle, Payload: mustJSON(protocol.Lifecycle{
+		Kind:       protocol.LifecycleStateChanged,
+		ProjectKey: s.projectKey,
+		ChildID:    id,
+		State:      string(state),
+	})})
+}
+
+func (s *clientSubscriber) OnPTYOut(childID string, chunk []byte) {
+	cp := append([]byte(nil), chunk...)
+	var size protocol.Size
+	var ownerID string
+	if s.project != nil {
+		size, ownerID, _ = s.project.PaneDisplay(childID)
+	}
+	f, err := protocol.NewFrame(protocol.FramePaneChunk, protocol.PaneChunk{PaneID: childID, Bytes: cp, Size: size, DisplayOwner: ownerID == "" || ownerID == s.clientID})
+	if err != nil {
+		return
+	}
+	select {
+	case s.frames <- f:
+	default:
+		s.mu.Lock()
+		s.snapshotRequired[childID] = true
+		s.mu.Unlock()
+	}
+}
+
+func (s *clientSubscriber) sendLifecycle(kind protocol.LifecycleKind, c *Child, state string) {
+	var child json.RawMessage
+	if c != nil {
+		child = mustJSON(serializeChildModel(c))
+	}
+	childID := ""
+	if c != nil {
+		childID = c.ID
+	}
+	s.sendFrame(protocol.Frame{Type: protocol.FrameLifecycle, Payload: mustJSON(protocol.Lifecycle{
+		Kind:       kind,
+		ProjectKey: s.projectKey,
+		ChildID:    childID,
+		Child:      child,
+		State:      state,
+	})})
+}
+
+func (s *clientSubscriber) sendFrame(f protocol.Frame) {
+	select {
+	case s.frames <- f:
+	default:
+		s.mu.Lock()
+		s.lifecycleDirty = true
+		s.mu.Unlock()
+	}
+}
+
+func mustJSON(v any) json.RawMessage {
+	b, err := json.Marshal(v)
+	if err != nil {
+		return nil
+	}
+	return b
+}

internal/app/client_subscriber_test.go

@@ -0,0 +1,32 @@
+package app
+
+import (
+	"testing"
+
+	"github.com/hjbdev/patterm/internal/protocol"
+)
+
+func TestClientSubscriberCopiesChunksAndMarksSnapshotOnOverflow(t *testing.T) {
+	sub := newClientSubscriber(&Project{Key: "project"}, "client", 1)
+	chunk := []byte("first")
+	sub.OnPTYOut("p_123456", chunk)
+	chunk[0] = 'X'
+
+	f, ok := sub.Recv()
+	if !ok {
+		t.Fatalf("Recv closed")
+	}
+	payload, err := protocol.Decode[protocol.PaneChunk](f)
+	if err != nil {
+		t.Fatalf("Decode: %v", err)
+	}
+	if string(payload.Bytes) != "first" {
+		t.Fatalf("payload retained pump buffer: %q", string(payload.Bytes))
+	}
+
+	sub.OnPTYOut("p_123456", []byte("queued"))
+	sub.OnPTYOut("p_123456", []byte("dropped"))
+	if !sub.SnapshotRequired("p_123456") {
+		t.Fatalf("overflow did not mark pane snapshot required")
+	}
+}

internal/app/client_view.go

@@ -0,0 +1,40 @@
+package app
+
+// ClientView is the per-client UI cursor over daemon-owned project/process
+// state. In loopback mode there is one view, owned by uiState; future network
+// clients will each get their own copy.
+type ClientView struct {
+	ID            string
+	ProjectKey    string
+	ProjectName   string
+	FocusedID     string
+	FocusedPad    string
+	ActiveAgentID string
+	PadOffset     int
+	PadOffsetName string
+	Cols          uint16
+	Rows          uint16
+}
+
+func (v *ClientView) FocusChild(id string) {
+	v.FocusedID = id
+	v.FocusedPad = ""
+}
+
+func (v *ClientView) FocusPad(name string) {
+	v.FocusedID = ""
+	v.FocusedPad = name
+	if v.PadOffsetName != name {
+		v.PadOffset = 0
+		v.PadOffsetName = name
+	}
+}
+
+func (v *ClientView) ClearPadFocus() {
+	v.FocusedPad = ""
+}
+
+func (v *ClientView) Resize(cols, rows uint16) {
+	v.Cols = cols
+	v.Rows = rows
+}

internal/app/cursorshift.go

@@ -67,6 +67,29 @@ func (cs *cursorShifter) clampCol(col int) int {
 	return col
 }
 
+// clampHostRow returns a host-coordinate row clamped to the viewport
+// rows mainTop..mainBottom. A child whose internal row state drifted
+// past the viewport (long-running claude / codex sessions) can issue a
+// CUP / HVP / VPA aimed at row hostRows; after the +rowOffset shift the
+// raw host target sits past the viewport bottom (the status row) or
+// above the viewport top (the tab bar). Without clamping the host
+// cursor lands on the chrome and the next printable wipes it. childRows
+// == 0 (uninitialised shifter, only seen in tests) disables clamping.
+func (cs *cursorShifter) clampHostRow(r int) int {
+	if cs.childRows <= 0 {
+		return r
+	}
+	minR := cs.rowOffset + 1
+	maxR := cs.rowOffset + cs.childRows
+	if r < minR {
+		return minR
+	}
+	if r > maxR {
+		return maxR
+	}
+	return r
+}
+
 // Shift consumes a chunk of PTY-master bytes, applies row offsets to
 // any complete CUP/HVP/VPA/DECSTBM sequences, and returns the rewritten
 // bytes. Partial sequences are buffered across calls so a CSI that
@@ -206,7 +229,7 @@ func (cs *cursorShifter) emitCSI() {
 			cs.pending.Write(cs.buf)
 			return
 		}
-		r += cs.rowOffset
+		r = cs.clampHostRow(r + cs.rowOffset)
 		c = cs.clampCol(c)
 		cs.pending.WriteString("\x1b[")
 		cs.pending.WriteString(strconv.Itoa(r))
@@ -226,13 +249,14 @@ func (cs *cursorShifter) emitCSI() {
 		cs.pending.WriteString(strconv.Itoa(c))
 		cs.pending.WriteByte(final)
 	case 'd':
-		// VPA: row.
+		// VPA: row. Clamp to the viewport so a child that drifted
+		// past its row count can't land the host cursor on the status row.
 		r, ok := parseOneParam(paramsRaw, 1)
 		if !ok {
 			cs.pending.Write(cs.buf)
 			return
 		}
-		r += cs.rowOffset
+		r = cs.clampHostRow(r + cs.rowOffset)
 		cs.pending.WriteString("\x1b[")
 		cs.pending.WriteString(strconv.Itoa(r))
 		cs.pending.WriteByte(final)

internal/app/cursorshift_test.go

@@ -111,3 +111,40 @@ func TestCursorShifterCUPNoClampWhenChildColsZero(t *testing.T) {
 		t.Fatalf("childCols=0 should disable col clamping: got %q", got)
 	}
 }
+
+// In longer claude sessions the cursor's internal row state could drift
+// past the viewport height. CUP / HVP / VPA without row clamping would
+// then land the host cursor on the status row or above the tab bar,
+// where the next printable wipes the chrome.
+func TestCursorShifterClampsCUPRowToMainBottom(t *testing.T) {
+	// rowOffset=2 (mainTop=3), childRows=36 → mainBottom=38.
+	cs := newCursorShifter(2, 36, 80)
+	got := cs.Shift([]byte("\x1b[40;5H"))
+	if string(got) != "\x1b[38;5H" {
+		t.Fatalf("CUP row 40 (post-shift 42) should clamp to 38: got %q", got)
+	}
+}
+
+func TestCursorShifterClampsHVPRowToMainBottom(t *testing.T) {
+	cs := newCursorShifter(2, 36, 80)
+	got := cs.Shift([]byte("\x1b[99;1f"))
+	if string(got) != "\x1b[38;1f" {
+		t.Fatalf("HVP row 99 should clamp to mainBottom: got %q", got)
+	}
+}
+
+func TestCursorShifterClampsVPARow(t *testing.T) {
+	cs := newCursorShifter(2, 36, 80)
+	got := cs.Shift([]byte("\x1b[60d"))
+	if string(got) != "\x1b[38d" {
+		t.Fatalf("VPA row 60 should clamp to mainBottom: got %q", got)
+	}
+}
+
+func TestCursorShifterCUPRowNoClampWhenChildRowsZero(t *testing.T) {
+	cs := newCursorShifter(2, 0, 80)
+	got := cs.Shift([]byte("\x1b[40;5H"))
+	if string(got) != "\x1b[42;5H" {
+		t.Fatalf("childRows=0 should disable row clamping: got %q", got)
+	}
+}

internal/app/daemon_core.go

@@ -0,0 +1,530 @@
+package app
+
+import (
+	"context"
+	"fmt"
+	"path/filepath"
+	"sort"
+	"sync"
+	"syscall"
+	"time"
+
+	"github.com/hjbdev/patterm/internal/mcp"
+	"github.com/hjbdev/patterm/internal/persist"
+	"github.com/hjbdev/patterm/internal/preset"
+	"github.com/hjbdev/patterm/internal/projectkey"
+	"github.com/hjbdev/patterm/internal/protocol"
+	"github.com/hjbdev/patterm/internal/scratchpad"
+	"github.com/hjbdev/patterm/internal/trust"
+)
+
+type Project struct {
+	Key  string
+	Dir  string
+	Name string
+
+	Session      *Session
+	Pads         *scratchpad.Store
+	Trust        *trust.Store
+	Persist      *persist.Store
+	Launcher     *Launcher
+	Host         *toolHost
+	savedProcess []persist.Entry
+
+	displayMu     sync.Mutex
+	displayOwners map[string]paneDisplayOwner
+
+	lastActive time.Time
+}
+
+type paneDisplayOwner struct {
+	ClientID string
+	Size     protocol.Size
+}
+
+type projectSummary struct {
+	Key       string
+	Dir       string
+	Name      string
+	TabCount  int
+	IsCurrent bool
+}
+
+// ProjectRegistry is the daemon-owned project map. Phase 1 still runs in one
+// local process, but every project already has isolated stores, session,
+// launcher, and tool host so future clients can attach to different projects.
+type ProjectRegistry struct {
+	mu       sync.Mutex
+	projects map[string]*Project
+
+	defaultProjectKey string
+	presets           preset.Set
+	settings          settings
+	mcpSrv            *mcp.Server
+	cols, rows        uint16
+}
+
+func newProjectRegistry(presets preset.Set, settings settings, mcpSrv *mcp.Server, cols, rows uint16) *ProjectRegistry {
+	return &ProjectRegistry{
+		projects: make(map[string]*Project),
+		presets:  presets,
+		settings: settings,
+		mcpSrv:   mcpSrv,
+		cols:     cols,
+		rows:     rows,
+	}
+}
+
+func (r *ProjectRegistry) Open(ctx context.Context, dir string) (*Project, error) {
+	key, err := projectkey.Key(dir)
+	if err != nil {
+		return nil, err
+	}
+	abs, err := filepath.Abs(dir)
+	if err != nil {
+		return nil, err
+	}
+
+	r.mu.Lock()
+	if p := r.projects[key]; p != nil {
+		p.lastActive = time.Now()
+		r.mu.Unlock()
+		return p, nil
+	}
+	r.mu.Unlock()
+
+	pads, err := scratchpad.Open(key)
+	if err != nil {
+		return nil, fmt.Errorf("app: scratchpad init: %w", err)
+	}
+	trustStore, err := trust.Open(key)
+	if err != nil {
+		return nil, fmt.Errorf("app: trust init: %w", err)
+	}
+	persistStore, err := persist.Open(key)
+	if err != nil {
+		return nil, fmt.Errorf("app: persist init: %w", err)
+	}
+	sess := NewSession(abs, key)
+	savedProcesses := persistStore.List()
+	for _, e := range savedProcesses {
+		_ = persistStore.Remove(e.ID)
+	}
+	sess.SetPersistStore(persistStore)
+	socket := ""
+	if r.mcpSrv != nil {
+		socket = r.mcpSrv.Socket()
+	}
+	launcher := NewLauncher(sess, socket, r.cols, r.rows)
+	host := newToolHost(sess, pads, launcher, r.presets, trustStore, r.cols, r.rows)
+	go sess.runClassifier(ctx)
+
+	p := &Project{
+		Key:           key,
+		Dir:           abs,
+		Name:          filepath.Base(abs),
+		Session:       sess,
+		Pads:          pads,
+		Trust:         trustStore,
+		Persist:       persistStore,
+		Launcher:      launcher,
+		Host:          host,
+		savedProcess:  savedProcesses,
+		displayOwners: make(map[string]paneDisplayOwner),
+		lastActive:    time.Now(),
+	}
+
+	r.mu.Lock()
+	if existing := r.projects[key]; existing != nil {
+		r.mu.Unlock()
+		sess.Shutdown()
+		return existing, nil
+	}
+	r.projects[key] = p
+	if r.defaultProjectKey == "" {
+		r.defaultProjectKey = key
+	}
+	r.mu.Unlock()
+	return p, nil
+}
+
+func (r *ProjectRegistry) Project(key string) *Project {
+	r.mu.Lock()
+	defer r.mu.Unlock()
+	return r.projects[key]
+}
+
+func (r *ProjectRegistry) Count() int {
+	r.mu.Lock()
+	defer r.mu.Unlock()
+	return len(r.projects)
+}
+
+func (r *ProjectRegistry) DefaultProject() *Project {
+	r.mu.Lock()
+	defer r.mu.Unlock()
+	return r.projects[r.defaultProjectKey]
+}
+
+func (p *Project) ClaimPaneDisplay(clientID, paneID string, size protocol.Size) (protocol.Size, bool) {
+	if p == nil || paneID == "" {
+		return size, true
+	}
+	if size.Cols == 0 || size.Rows == 0 {
+		size = protocol.Size{Cols: 80, Rows: 24}
+	}
+	p.displayMu.Lock()
+	if p.displayOwners == nil {
+		p.displayOwners = make(map[string]paneDisplayOwner)
+	}
+	owner, ok := p.displayOwners[paneID]
+	if !ok || owner.ClientID == "" || owner.ClientID == clientID {
+		p.displayOwners[paneID] = paneDisplayOwner{ClientID: clientID, Size: size}
+		p.displayMu.Unlock()
+		p.Session.ResizeChild(paneID, size.Cols, size.Rows)
+		return size, true
+	}
+	p.displayMu.Unlock()
+	return owner.Size, false
+}
+
+func (p *Project) ResizeClientDisplays(clientID string, size protocol.Size) {
+	if p == nil || size.Cols == 0 || size.Rows == 0 {
+		return
+	}
+	p.displayMu.Lock()
+	var panes []string
+	for paneID, owner := range p.displayOwners {
+		if owner.ClientID != clientID {
+			continue
+		}
+		owner.Size = size
+		p.displayOwners[paneID] = owner
+		panes = append(panes, paneID)
+	}
+	p.displayMu.Unlock()
+	for _, paneID := range panes {
+		p.Session.ResizeChild(paneID, size.Cols, size.Rows)
+	}
+	p.Launcher.SetSize(size.Cols, size.Rows)
+	p.Host.SetSize(size.Cols, size.Rows)
+}
+
+func (p *Project) ReleaseClientDisplays(clientID string) {
+	if p == nil {
+		return
+	}
+	p.displayMu.Lock()
+	for paneID, owner := range p.displayOwners {
+		if owner.ClientID == clientID {
+			delete(p.displayOwners, paneID)
+		}
+	}
+	p.displayMu.Unlock()
+}
+
+func (p *Project) PaneDisplay(paneID string) (protocol.Size, string, bool) {
+	if p == nil || paneID == "" {
+		return protocol.Size{}, "", false
+	}
+	p.displayMu.Lock()
+	defer p.displayMu.Unlock()
+	owner, ok := p.displayOwners[paneID]
+	return owner.Size, owner.ClientID, ok
+}
+
+func (r *ProjectRegistry) Shutdown() {
+	r.mu.Lock()
+	projects := make([]*Project, 0, len(r.projects))
+	for _, p := range r.projects {
+		projects = append(projects, p)
+	}
+	r.mu.Unlock()
+	for _, p := range projects {
+		p.Session.Shutdown()
+	}
+}
+
+func (r *ProjectRegistry) ResizeAll(cols, rows uint16) {
+	r.mu.Lock()
+	r.cols, r.rows = cols, rows
+	projects := make([]*Project, 0, len(r.projects))
+	for _, p := range r.projects {
+		projects = append(projects, p)
+	}
+	r.mu.Unlock()
+	for _, p := range projects {
+		p.Session.ResizeAll(cols, rows)
+		p.Launcher.SetSize(cols, rows)
+		p.Host.SetSize(cols, rows)
+	}
+}
+
+func (r *ProjectRegistry) Summaries(currentKey string) []projectSummary {
+	r.mu.Lock()
+	defer r.mu.Unlock()
+	out := make([]projectSummary, 0, len(r.projects))
+	for _, p := range r.projects {
+		out = append(out, projectSummary{
+			Key:       p.Key,
+			Dir:       p.Dir,
+			Name:      p.Name,
+			TabCount:  len(runningTopLevels(p.Session.Children())),
+			IsCurrent: p.Key == currentKey,
+		})
+	}
+	sort.Slice(out, func(i, j int) bool {
+		if out[i].IsCurrent != out[j].IsCurrent {
+			return out[i].IsCurrent
+		}
+		return out[i].Name < out[j].Name
+	})
+	return out
+}
+
+func (r *ProjectRegistry) findProjectByChild(id string) (*Project, *Child) {
+	if id == "" {
+		return nil, nil
+	}
+	r.mu.Lock()
+	projects := make([]*Project, 0, len(r.projects))
+	for _, p := range r.projects {
+		projects = append(projects, p)
+	}
+	r.mu.Unlock()
+	for _, p := range projects {
+		if c := p.Session.FindChild(id); c != nil {
+			return p, c
+		}
+	}
+	return nil, nil
+}
+
+func (r *ProjectRegistry) projectForCaller(callerID string) *Project {
+	if p, _ := r.findProjectByChild(callerID); p != nil {
+		return p
+	}
+	r.mu.Lock()
+	defer r.mu.Unlock()
+	return r.projects[r.defaultProjectKey]
+}
+
+func (r *ProjectRegistry) hostForCaller(callerID string) *toolHost {
+	if p := r.projectForCaller(callerID); p != nil {
+		return p.Host
+	}
+	return nil
+}
+
+func (r *ProjectRegistry) hostForProcess(processID string) *toolHost {
+	if p, _ := r.findProjectByChild(processID); p != nil {
+		return p.Host
+	}
+	return nil
+}
+
+func (r *ProjectRegistry) ResolveCallerIdentity(identity string) string {
+	r.mu.Lock()
+	projects := make([]*Project, 0, len(r.projects))
+	for _, p := range r.projects {
+		projects = append(projects, p)
+	}
+	r.mu.Unlock()
+	for _, p := range projects {
+		if c := p.Session.FindChildByIdentity(identity); c != nil {
+			return c.ID
+		}
+	}
+	return ""
+}
+
+func (r *ProjectRegistry) CallerRole(processID string) mcp.CallerRole {
+	if h := r.hostForCaller(processID); h != nil {
+		return h.CallerRole(processID)
+	}
+	return mcp.RoleOrchestrator
+}
+
+func (r *ProjectRegistry) SpawnAgent(callerID string, args mcp.SpawnAgentArgs) (mcp.ProcessInfo, error) {
+	return r.hostForCaller(callerID).SpawnAgent(callerID, args)
+}
+
+func (r *ProjectRegistry) SpawnProcess(callerID string, args mcp.SpawnProcessArgs) (mcp.ProcessInfo, error) {
+	return r.hostForCaller(callerID).SpawnProcess(callerID, args)
+}
+
+func (r *ProjectRegistry) StartProcess(callerID, processID string) (mcp.ProcessInfo, error) {
+	if h := r.hostForProcess(processID); h != nil {
+		return h.StartProcess(callerID, processID)
+	}
+	return mcp.ProcessInfo{}, mcp.Errorf(mcp.ErrorKindNotFound, "no such process %q", processID)
+}
+
+func (r *ProjectRegistry) RestartProcess(callerID, processID string, sig syscall.Signal) (mcp.ProcessInfo, error) {
+	if h := r.hostForProcess(processID); h != nil {
+		return h.RestartProcess(callerID, processID, sig)
+	}
+	return mcp.ProcessInfo{}, mcp.Errorf(mcp.ErrorKindNotFound, "no such process %q", processID)
+}
+
+func (r *ProjectRegistry) StopProcess(callerID, processID string, sig syscall.Signal) (mcp.ProcessInfo, error) {
+	if h := r.hostForProcess(processID); h != nil {
+		return h.StopProcess(callerID, processID, sig)
+	}
+	return mcp.ProcessInfo{}, mcp.Errorf(mcp.ErrorKindNotFound, "no such process %q", processID)
+}
+
+func (r *ProjectRegistry) CloseProcess(callerID, processID string) error {
+	if h := r.hostForProcess(processID); h != nil {
+		return h.CloseProcess(callerID, processID)
+	}
+	return mcp.Errorf(mcp.ErrorKindNotFound, "no such process %q", processID)
+}
+
+func (r *ProjectRegistry) RenameProcess(callerID, processID, name string) error {
+	if h := r.hostForProcess(processID); h != nil {
+		return h.RenameProcess(callerID, processID, name)
+	}
+	return mcp.Errorf(mcp.ErrorKindNotFound, "no such process %q", processID)
+}
+
+func (r *ProjectRegistry) SelectProcess(callerID, processID string) error {
+	if h := r.hostForProcess(processID); h != nil {
+		return h.SelectProcess(callerID, processID)
+	}
+	return mcp.Errorf(mcp.ErrorKindNotFound, "no such process %q", processID)
+}
+
+func (r *ProjectRegistry) ListProcesses(callerID, kindFilter string) []mcp.ProcessInfo {
+	if h := r.hostForCaller(callerID); h != nil {
+		return h.ListProcesses(callerID, kindFilter)
+	}
+	return nil
+}
+
+func (r *ProjectRegistry) GetProcessStatus(callerID, processID string) (mcp.ProcessStatus, error) {
+	if h := r.hostForProcess(processID); h != nil {
+		return h.GetProcessStatus(callerID, processID)
+	}
+	return mcp.ProcessStatus{}, mcp.Errorf(mcp.ErrorKindNotFound, "no such process %q", processID)
+}
+
+func (r *ProjectRegistry) GetProjectStatus(callerID string) (mcp.ProjectStatus, error) {
+	return r.hostForCaller(callerID).GetProjectStatus(callerID)
+}
+
+func (r *ProjectRegistry) GetProcessOutput(callerID, processID, mode string, sinceOffset int64) (mcp.ProcessOutput, error) {
+	if h := r.hostForProcess(processID); h != nil {
+		return h.GetProcessOutput(callerID, processID, mode, sinceOffset)
+	}
+	return mcp.ProcessOutput{}, mcp.Errorf(mcp.ErrorKindNotFound, "no such process %q", processID)
+}
+
+func (r *ProjectRegistry) GetProcessRawOutput(callerID, processID string, sinceOffset int64) (mcp.RawOutput, error) {
+	if h := r.hostForProcess(processID); h != nil {
+		return h.GetProcessRawOutput(callerID, processID, sinceOffset)
+	}
+	return mcp.RawOutput{}, mcp.Errorf(mcp.ErrorKindNotFound, "no such process %q", processID)
+}
+
+func (r *ProjectRegistry) SearchOutput(callerID, processID, pattern, kind string, limit int) (mcp.SearchResult, error) {
+	if h := r.hostForProcess(processID); h != nil {
+		return h.SearchOutput(callerID, processID, pattern, kind, limit)
+	}
+	return mcp.SearchResult{}, mcp.Errorf(mcp.ErrorKindNotFound, "no such process %q", processID)
+}
+
+func (r *ProjectRegistry) WaitForPattern(callerID, processID, pattern string, timeoutSeconds float64, scope string) (bool, string, error) {
+	if h := r.hostForProcess(processID); h != nil {
+		return h.WaitForPattern(callerID, processID, pattern, timeoutSeconds, scope)
+	}
+	return false, "", mcp.Errorf(mcp.ErrorKindNotFound, "no such process %q", processID)
+}
+
+func (r *ProjectRegistry) GetProcessPorts(callerID, processID string) ([]mcp.PortSighting, error) {
+	if h := r.hostForProcess(processID); h != nil {
+		return h.GetProcessPorts(callerID, processID)
+	}
+	return nil, mcp.Errorf(mcp.ErrorKindNotFound, "no such process %q", processID)
+}
+
+func (r *ProjectRegistry) SendInput(callerID string, args mcp.SendInputArgs) (mcp.SendInputResult, error) {
+	if h := r.hostForProcess(args.ProcessID); h != nil {
+		return h.SendInput(callerID, args)
+	}
+	return mcp.SendInputResult{}, mcp.Errorf(mcp.ErrorKindNotFound, "no such process %q", args.ProcessID)
+}
+
+func (r *ProjectRegistry) SendMessage(callerID, targetID, message string) error {
+	if h := r.hostForProcess(targetID); h != nil {
+		return h.SendMessage(callerID, targetID, message)
+	}
+	return mcp.Errorf(mcp.ErrorKindNotFound, "no such process %q", targetID)
+}
+
+func (r *ProjectRegistry) RequestHumanAttention(callerID, processID, reason string) error {
+	if h := r.hostForProcess(processID); h != nil {
+		return h.RequestHumanAttention(callerID, processID, reason)
+	}
+	return mcp.Errorf(mcp.ErrorKindNotFound, "no such process %q", processID)
+}
+
+func (r *ProjectRegistry) TimerWait(callerID string, seconds float64, label string) (string, error) {
+	return r.hostForCaller(callerID).TimerWait(callerID, seconds, label)
+}
+
+func (r *ProjectRegistry) TimerSet(callerID string, args mcp.TimerSetArgs) (mcp.TimerHandle, error) {
+	return r.hostForCaller(callerID).TimerSet(callerID, args)
+}
+
+func (r *ProjectRegistry) TimerFireWhenIdleAny(callerID string, args mcp.TimerFireWhenIdleArgs) (mcp.TimerFireWhenIdleResponse, error) {
+	return r.hostForCaller(callerID).TimerFireWhenIdleAny(callerID, args)
+}
+
+func (r *ProjectRegistry) TimerFireWhenIdleAll(callerID string, args mcp.TimerFireWhenIdleArgs) (mcp.TimerFireWhenIdleResponse, error) {
+	return r.hostForCaller(callerID).TimerFireWhenIdleAll(callerID, args)
+}
+
+func (r *ProjectRegistry) TimerCancel(callerID, id string) error {
+	return r.hostForCaller(callerID).TimerCancel(callerID, id)
+}
+
+func (r *ProjectRegistry) TimerPause(callerID, id string) error {
+	return r.hostForCaller(callerID).TimerPause(callerID, id)
+}
+
+func (r *ProjectRegistry) TimerResume(callerID, id string) error {
+	return r.hostForCaller(callerID).TimerResume(callerID, id)
+}
+
+func (r *ProjectRegistry) TimerList(callerID string) ([]mcp.TimerInfo, error) {
+	return r.hostForCaller(callerID).TimerList(callerID)
+}
+
+func (r *ProjectRegistry) ScratchpadList(callerID string) ([]scratchpad.Entry, error) {
+	return r.hostForCaller(callerID).ScratchpadList(callerID)
+}
+
+func (r *ProjectRegistry) ScratchpadRead(callerID, name string) (string, string, error) {
+	return r.hostForCaller(callerID).ScratchpadRead(callerID, name)
+}
+
+func (r *ProjectRegistry) ScratchpadWrite(callerID, name, content, expectedRevision string) (string, error) {
+	return r.hostForCaller(callerID).ScratchpadWrite(callerID, name, content, expectedRevision)
+}
+
+func (r *ProjectRegistry) ScratchpadAppend(callerID, name, content string) error {
+	return r.hostForCaller(callerID).ScratchpadAppend(callerID, name, content)
+}
+
+func (r *ProjectRegistry) ScratchpadDelete(callerID, name string) error {
+	return r.hostForCaller(callerID).ScratchpadDelete(callerID, name)
+}
+
+func (r *ProjectRegistry) WhoAmI(callerID string) mcp.WhoAmI {
+	return r.hostForCaller(callerID).WhoAmI(callerID)
+}
+
+func (r *ProjectRegistry) Help(callerID, topic string) mcp.HelpResponse {
+	return r.hostForCaller(callerID).Help(callerID, topic)
+}

internal/app/daemon_net.go

@@ -0,0 +1,481 @@
+package app
+
+import (
+	"context"
+	"encoding/json"
+	"errors"
+	"fmt"
+	"io"
+	"net"
+	"os"
+	"path/filepath"
+	"strconv"
+	"strings"
+	"sync"
+	"syscall"
+	"time"
+
+	"github.com/hjbdev/patterm/internal/mcp"
+	"github.com/hjbdev/patterm/internal/preset"
+	"github.com/hjbdev/patterm/internal/protocol"
+)
+
+type DaemonOptions struct {
+	ProjectDir  string
+	SocketPath  string
+	PidPath     string
+	ListenAddr  string
+	Token       string
+	TokenOut    io.Writer
+	ListenReady chan string
+	Cols        uint16
+	Rows        uint16
+}
+
+type DaemonStatus struct {
+	PID      int
+	Socket   string
+	Projects []protocol.Project
+}
+
+func RuntimeDaemonPaths() (socketPath, pidPath string, err error) {
+	base := os.Getenv("XDG_RUNTIME_DIR")
+	if base == "" {
+		base = os.TempDir()
+	}
+	dir := filepath.Join(base, "patterm")
+	if err := os.MkdirAll(dir, 0o700); err != nil {
+		return "", "", err
+	}
+	return filepath.Join(dir, "daemon.sock"), filepath.Join(dir, "daemon.pid"), nil
+}
+
+func RunDaemon(ctx context.Context, opts DaemonOptions) error {
+	if opts.ProjectDir == "" {
+		cwd, err := os.Getwd()
+		if err != nil {
+			return err
+		}
+		opts.ProjectDir = cwd
+	}
+	if opts.SocketPath == "" || opts.PidPath == "" {
+		socket, pid, err := RuntimeDaemonPaths()
+		if err != nil {
+			return err
+		}
+		if opts.SocketPath == "" {
+			opts.SocketPath = socket
+		}
+		if opts.PidPath == "" {
+			opts.PidPath = pid
+		}
+	}
+	if opts.Cols == 0 {
+		opts.Cols = 80
+	}
+	if opts.Rows == 0 {
+		opts.Rows = 24
+	}
+	lockPath, err := prepareDaemonSocket(opts.SocketPath, opts.PidPath)
+	if err != nil {
+		return err
+	}
+	defer os.Remove(lockPath)
+	ln, err := net.Listen("unix", opts.SocketPath)
+	if err != nil {
+		return fmt.Errorf("daemon: listen %s: %w", opts.SocketPath, err)
+	}
+	defer ln.Close()
+	defer os.Remove(opts.SocketPath)
+	if err := os.Chmod(opts.SocketPath, 0o600); err != nil {
+		return err
+	}
+	if err := os.WriteFile(opts.PidPath, []byte(strconv.Itoa(os.Getpid())+"\n"), 0o600); err != nil {
+		return err
+	}
+	defer os.Remove(opts.PidPath)
+
+	presets, err := preset.Load()
+	if err != nil {
+		return fmt.Errorf("daemon: load presets: %w", err)
+	}
+	appSettings, _, err := loadSettings()
+	if err != nil {
+		logf("daemon settings load: %v", err)
+	}
+	mcpSrv, err := mcp.Start()
+	if err != nil {
+		return fmt.Errorf("daemon: mcp start: %w", err)
+	}
+	defer mcpSrv.Close()
+
+	ctx, cancel := context.WithCancel(ctx)
+	defer cancel()
+	registry := newProjectRegistry(presets, appSettings, mcpSrv, opts.Cols, opts.Rows)
+	defer registry.Shutdown()
+	mcpSrv.SetHost(registry)
+	if _, err := registry.Open(ctx, opts.ProjectDir); err != nil {
+		return err
+	}
+
+	var tcpLn net.Listener
+	tcpToken := opts.Token
+	if opts.ListenAddr != "" {
+		addr := normalizeListenAddr(opts.ListenAddr)
+		tcpToken, err = ensureDaemonToken(tcpToken)
+		if err != nil {
+			return err
+		}
+		tcpLn, err = net.Listen("tcp", addr)
+		if err != nil {
+			return fmt.Errorf("daemon: listen tcp %s: %w", addr, err)
+		}
+		defer tcpLn.Close()
+		if opts.ListenReady != nil {
+			select {
+			case opts.ListenReady <- tcpLn.Addr().String():
+			default:
+			}
+		}
+		out := opts.TokenOut
+		if out == nil {
+			out = os.Stderr
+		}
+		fmt.Fprintf(out, "patterm daemon listening on %s\npatterm token: %s\n", tcpLn.Addr().String(), tcpToken)
+	}
+
+	var wg sync.WaitGroup
+	go func() {
+		<-ctx.Done()
+		_ = ln.Close()
+		if tcpLn != nil {
+			_ = tcpLn.Close()
+		}
+	}()
+	errCh := make(chan error, 2)
+	go acceptDaemonLoop(ctx, &wg, ln, "", cancel, registry, errCh)
+	if tcpLn != nil {
+		go acceptDaemonLoop(ctx, &wg, tcpLn, tcpToken, cancel, registry, errCh)
+	}
+	select {
+	case <-ctx.Done():
+	case err := <-errCh:
+		cancel()
+		wg.Wait()
+		return err
+	}
+	wg.Wait()
+	return nil
+}
+
+func acceptDaemonLoop(ctx context.Context, wg *sync.WaitGroup, ln net.Listener, authToken string, stop func(), registry *ProjectRegistry, errCh chan<- error) {
+	for {
+		conn, err := ln.Accept()
+		if err != nil {
+			if errors.Is(err, net.ErrClosed) || ctx.Err() != nil {
+				return
+			}
+			select {
+			case errCh <- err:
+			default:
+			}
+			return
+		}
+		wg.Add(1)
+		go func() {
+			defer wg.Done()
+			handleDaemonConn(ctx, stop, registry, protocol.NewConnTransport(conn), authToken)
+		}()
+	}
+}
+
+func normalizeListenAddr(addr string) string {
+	addr = strings.TrimSpace(addr)
+	if addr == "" {
+		return ""
+	}
+	if _, _, err := net.SplitHostPort(addr); err == nil {
+		return addr
+	}
+	if strings.HasPrefix(addr, ":") {
+		return addr
+	}
+	if _, err := strconv.Atoi(addr); err == nil {
+		return ":" + addr
+	}
+	return addr
+}
+
+func ensureDaemonToken(token string) (string, error) {
+	if strings.TrimSpace(token) != "" {
+		return strings.TrimSpace(token), nil
+	}
+	return LoadOrCreateClientToken()
+}
+
+func prepareDaemonSocket(socketPath, pidPath string) (string, error) {
+	if err := os.MkdirAll(filepath.Dir(socketPath), 0o700); err != nil {
+		return "", err
+	}
+	lockPath := pidPath + ".lock"
+	if data, err := os.ReadFile(pidPath); err == nil {
+		if pid, perr := strconv.Atoi(strings.TrimSpace(string(data))); perr == nil && pid > 0 {
+			if sigErr := syscallSignal0(pid); sigErr == nil {
+				return "", fmt.Errorf("daemon already running with pid %d", pid)
+			}
+		}
+	}
+	_ = os.Remove(socketPath)
+	_ = os.Remove(pidPath)
+	_ = os.Remove(lockPath)
+	f, err := os.OpenFile(lockPath, os.O_CREATE|os.O_EXCL|os.O_WRONLY, 0o600)
+	if err != nil {
+		return "", fmt.Errorf("daemon: lock %s: %w", lockPath, err)
+	}
+	_, _ = f.WriteString(strconv.Itoa(os.Getpid()) + "\n")
+	_ = f.Close()
+	return lockPath, nil
+}
+
+func syscallSignal0(pid int) error {
+	return syscall.Kill(pid, 0)
+}
+
+func handleDaemonConn(ctx context.Context, stop func(), registry *ProjectRegistry, t protocol.Transport, authToken string) {
+	defer t.Close()
+	f, err := t.Recv()
+	if err != nil {
+		return
+	}
+	switch f.Type {
+	case protocol.FrameList:
+		_ = sendProjectList(t, registry, "")
+		return
+	case protocol.FrameStop:
+		_ = sendProjectList(t, registry, "")
+		stop()
+		return
+	case protocol.FrameAttach:
+		if authToken != "" {
+			attach, err := protocol.Decode[protocol.Attach](f)
+			if err != nil {
+				_ = sendProtocolError(t, err.Error())
+				return
+			}
+			if attach.Token != authToken {
+				_ = sendProtocolError(t, "auth denied")
+				return
+			}
+		}
+		handleDaemonAttach(ctx, registry, t, f)
+	default:
+		_ = sendProtocolError(t, fmt.Sprintf("first frame must be attach, list, or stop; got %q", f.Type))
+	}
+}
+
+func handleDaemonAttach(ctx context.Context, registry *ProjectRegistry, t protocol.Transport, first protocol.Frame) {
+	attach, err := protocol.Decode[protocol.Attach](first)
+	if err != nil {
+		_ = sendProtocolError(t, err.Error())
+		return
+	}
+	project := registry.Project(attach.ProjectKey)
+	if project == nil && attach.ProjectPath != "" {
+		project, err = registry.Open(ctx, attach.ProjectPath)
+		if err != nil {
+			_ = sendProtocolError(t, err.Error())
+			return
+		}
+	}
+	if project == nil {
+		project = registry.DefaultProject()
+	}
+	if project == nil {
+		_ = sendProtocolError(t, "no project open")
+		return
+	}
+	clientID := fmt.Sprintf("c-%d", time.Now().UnixNano())
+	view := ClientView{
+		ID:          clientID,
+		ProjectKey:  project.Key,
+		ProjectName: project.Name,
+		Cols:        attach.TermSize.Cols,
+		Rows:        attach.TermSize.Rows,
+	}
+	if child := firstRunningTopLevel(project.Session.Children()); child != nil {
+		view.FocusChild(child.ID)
+		project.ClaimPaneDisplay(clientID, child.ID, attach.TermSize)
+	}
+	sub := newClientSubscriber(project, clientID, defaultClientSubscriberQueue)
+	project.Session.SubscribeClient(sub)
+	defer project.Session.UnsubscribeClient(sub)
+	defer project.ReleaseClientDisplays(clientID)
+
+	_ = sendHello(t, project, view.ID)
+	_ = sendProjectList(t, registry, project.Key)
+	_ = sendChrome(t, project, view)
+	if view.FocusedID != "" {
+		_ = sendSnapshot(t, project, clientID, view.FocusedID)
+	}
+
+	// Close the transport when the daemon context is cancelled (shutdown or
+	// `daemon stop`). Without this the t.Recv() loop below blocks forever on a
+	// still-connected client and the accept loop's wg.Wait() never returns.
+	go func() {
+		<-ctx.Done()
+		_ = t.Close()
+	}()
+
+	done := make(chan struct{})
+	go func() {
+		defer close(done)
+		for {
+			f, ok := sub.Recv()
+			if !ok {
+				return
+			}
+			if err := t.Send(f); err != nil {
+				return
+			}
+		}
+	}()
+
+	for {
+		f, err := t.Recv()
+		if err != nil {
+			return
+		}
+		switch f.Type {
+		case protocol.FrameDetach:
+			return
+		case protocol.FrameInput:
+			msg, err := protocol.Decode[protocol.Input](f)
+			if err == nil {
+				if c := project.Session.FindChild(msg.PaneID); c != nil {
+					_ = c.InjectAsUser(msg.Bytes)
+				}
+			}
+		case protocol.FrameResize:
+			msg, err := protocol.Decode[protocol.Resize](f)
+			if err == nil {
+				view.Resize(msg.Size.Cols, msg.Size.Rows)
+				if view.FocusedID != "" {
+					if _, _, ok := project.PaneDisplay(view.FocusedID); !ok {
+						project.ClaimPaneDisplay(clientID, view.FocusedID, msg.Size)
+					}
+				}
+				project.ResizeClientDisplays(clientID, msg.Size)
+			}
+		case protocol.FrameFocus:
+			msg, err := protocol.Decode[protocol.Focus](f)
+			if err == nil && msg.PaneID != "" {
+				view.FocusChild(msg.PaneID)
+				project.ClaimPaneDisplay(clientID, msg.PaneID, protocol.Size{Cols: view.Cols, Rows: view.Rows})
+				_ = sendChrome(t, project, view)
+				_ = sendSnapshot(t, project, clientID, msg.PaneID)
+			}
+		case protocol.FramePaletteCommand:
+			if child := handleDaemonPaletteCommand(project, f); child != nil {
+				view.FocusChild(child.ID)
+				project.ClaimPaneDisplay(clientID, child.ID, protocol.Size{Cols: view.Cols, Rows: view.Rows})
+				_ = sendChrome(t, project, view)
+				_ = sendSnapshot(t, project, clientID, child.ID)
+			}
+		}
+		select {
+		case <-done:
+			return
+		default:
+		}
+	}
+}
+
+func handleDaemonPaletteCommand(project *Project, f protocol.Frame) *Child {
+	msg, err := protocol.Decode[protocol.PaletteCommand](f)
+	if err != nil {
+		return nil
+	}
+	switch msg.Kind {
+	case "spawn_command":
+		var p struct {
+			Argv    []string `json:"argv"`
+			Name    string   `json:"name"`
+			WorkDir string   `json:"working_dir"`
+			Shell   bool     `json:"shell"`
+		}
+		if err := json.Unmarshal(msg.Data, &p); err != nil || len(p.Argv) == 0 {
+			return nil
+		}
+		name := p.Name
+		if name == "" {
+			name = strings.Join(p.Argv, " ")
+		}
+		c, err := project.Launcher.LaunchCommandArgv(p.Argv, name, "", p.WorkDir, nil, p.Shell)
+		if err != nil {
+			return nil
+		}
+		return c
+	}
+	return nil
+}
+
+func sendHello(t protocol.Transport, p *Project, clientID string) error {
+	f, err := protocol.NewFrame(protocol.FrameHello, protocol.Hello{Version: 1, DaemonID: strconv.Itoa(os.Getpid()), ClientID: clientID, ProjectKey: p.Key})
+	if err != nil {
+		return err
+	}
+	return t.Send(f)
+}
+
+func sendProjectList(t protocol.Transport, registry *ProjectRegistry, current string) error {
+	summaries := registry.Summaries(current)
+	projects := make([]protocol.Project, 0, len(summaries))
+	for _, p := range summaries {
+		projects = append(projects, protocol.Project{Key: p.Key, Path: p.Dir, Name: p.Name, TabCount: p.TabCount})
+	}
+	f, err := protocol.NewFrame(protocol.FrameProjectList, protocol.ProjectList{Projects: projects})
+	if err != nil {
+		return err
+	}
+	return t.Send(f)
+}
+
+func sendChrome(t protocol.Transport, p *Project, view ClientView) error {
+	pads, _ := p.Pads.List()
+	model := buildChromeModel(p.Key, view, p.Session.Children(), pads)
+	b, err := json.Marshal(model)
+	if err != nil {
+		return err
+	}
+	f, err := protocol.NewFrame(protocol.FrameChrome, protocol.Chrome{ProjectKey: p.Key, Model: b})
+	if err != nil {
+		return err
+	}
+	return t.Send(f)
+}
+
+func sendSnapshot(t protocol.Transport, p *Project, clientID, paneID string) error {
+	b, err := p.Session.SerializeChild(paneID)
+	if err != nil {
+		return nil
+	}
+	size, ownerID, _ := p.PaneDisplay(paneID)
+	f, err := protocol.NewFrame(protocol.FramePaneSnapshot, protocol.PaneSnapshot{
+		PaneID:       paneID,
+		Bytes:        b,
+		Size:         size,
+		DisplayOwner: ownerID == "" || ownerID == clientID,
+	})
+	if err != nil {
+		return err
+	}
+	return t.Send(f)
+}
+
+func sendProtocolError(t protocol.Transport, msg string) error {
+	f, err := protocol.NewFrame(protocol.FrameError, protocol.Error{Message: msg})
+	if err != nil {
+		return err
+	}
+	return t.Send(f)
+}

internal/app/daemon_net_test.go

@@ -0,0 +1,477 @@
+package app
+
+import (
+	"context"
+	"encoding/json"
+	"io"
+	"net"
+	"os"
+	"path/filepath"
+	"strings"
+	"testing"
+	"time"
+
+	"github.com/hjbdev/patterm/internal/preset"
+	"github.com/hjbdev/patterm/internal/protocol"
+)
+
+func TestDaemonDetachReattachPreservesProcess(t *testing.T) {
+	root := t.TempDir()
+	t.Setenv("XDG_CONFIG_HOME", filepath.Join(root, "config"))
+	t.Setenv("XDG_DATA_HOME", filepath.Join(root, "data"))
+	t.Setenv("XDG_RUNTIME_DIR", filepath.Join(root, "runtime"))
+	projectDir := filepath.Join(root, "project")
+	if err := os.MkdirAll(projectDir, 0o700); err != nil {
+		t.Fatalf("mkdir project: %v", err)
+	}
+	socket := filepath.Join(root, "runtime", "patterm", "daemon.sock")
+	pid := filepath.Join(root, "runtime", "patterm", "daemon.pid")
+
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+	errCh := make(chan error, 1)
+	go func() {
+		errCh <- RunDaemon(ctx, DaemonOptions{
+			ProjectDir: projectDir,
+			SocketPath: socket,
+			PidPath:    pid,
+			Cols:       80,
+			Rows:       24,
+		})
+	}()
+	waitForSocket(t, socket, errCh)
+
+	client1 := dialDaemon(t, socket)
+	sendFrame(t, client1, protocol.FrameAttach, protocol.Attach{
+		ProjectPath: projectDir,
+		TermSize:    protocol.Size{Cols: 80, Rows: 24},
+	})
+	expectFrame(t, client1, protocol.FrameHello)
+	expectFrame(t, client1, protocol.FrameProjectList)
+	expectFrame(t, client1, protocol.FrameChrome)
+
+	data, _ := json.Marshal(map[string]any{
+		"argv": []string{"sh", "-c", "trap 'exit 0' TERM; while :; do echo STILL-HERE; sleep 1; done"},
+		"name": "survivor",
+	})
+	sendFrame(t, client1, protocol.FramePaletteCommand, protocol.PaletteCommand{
+		Kind: "spawn_command",
+		Data: data,
+	})
+	waitForLifecycle(t, client1, protocol.LifecycleSpawned, 3*time.Second)
+	sendFrame(t, client1, protocol.FrameDetach, protocol.Detach{})
+	_ = client1.Close()
+
+	client2 := dialDaemon(t, socket)
+	defer client2.Close()
+	sendFrame(t, client2, protocol.FrameAttach, protocol.Attach{
+		ProjectPath: projectDir,
+		TermSize:    protocol.Size{Cols: 80, Rows: 24},
+	})
+	expectFrame(t, client2, protocol.FrameHello)
+	expectFrame(t, client2, protocol.FrameProjectList)
+	chrome := expectChrome(t, client2)
+	if !chromeHasProcess(chrome, "survivor") {
+		t.Fatalf("reattached chrome did not include surviving process: %s", string(chrome.Model))
+	}
+	expectFrame(t, client2, protocol.FramePaneSnapshot)
+
+	cancel()
+	select {
+	case err := <-errCh:
+		if err != nil {
+			t.Fatalf("daemon returned error: %v", err)
+		}
+	case <-time.After(3 * time.Second):
+		t.Fatalf("daemon did not stop")
+	}
+}
+
+func TestDaemonTCPTokenAuthAndUnixExemption(t *testing.T) {
+	root := t.TempDir()
+	t.Setenv("XDG_CONFIG_HOME", filepath.Join(root, "config"))
+	t.Setenv("XDG_DATA_HOME", filepath.Join(root, "data"))
+	t.Setenv("XDG_RUNTIME_DIR", filepath.Join(root, "runtime"))
+	projectDir := filepath.Join(root, "project")
+	if err := os.MkdirAll(projectDir, 0o700); err != nil {
+		t.Fatalf("mkdir project: %v", err)
+	}
+	socket := filepath.Join(root, "runtime", "patterm", "daemon.sock")
+	pid := filepath.Join(root, "runtime", "patterm", "daemon.pid")
+
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+	errCh := make(chan error, 1)
+	ready := make(chan string, 1)
+	go func() {
+		errCh <- RunDaemon(ctx, DaemonOptions{
+			ProjectDir:  projectDir,
+			SocketPath:  socket,
+			PidPath:     pid,
+			ListenAddr:  "127.0.0.1:0",
+			Token:       "secret-token",
+			TokenOut:    io.Discard,
+			ListenReady: ready,
+			Cols:        80,
+			Rows:        24,
+		})
+	}()
+	waitForSocket(t, socket, errCh)
+	tcpAddr := waitForTCPAddr(t, ready, errCh)
+
+	assertTCPAttachDenied(t, tcpAddr, "")
+	assertTCPAttachDenied(t, tcpAddr, "wrong-token")
+
+	tcpClient := dialTCPDaemon(t, tcpAddr)
+	defer tcpClient.Close()
+	sendFrame(t, tcpClient, protocol.FrameAttach, protocol.Attach{
+		Token:       "secret-token",
+		ProjectPath: projectDir,
+		TermSize:    protocol.Size{Cols: 80, Rows: 24},
+	})
+	expectFrame(t, tcpClient, protocol.FrameHello)
+	expectFrame(t, tcpClient, protocol.FrameProjectList)
+	expectFrame(t, tcpClient, protocol.FrameChrome)
+	data, _ := json.Marshal(map[string]any{
+		"argv": []string{"sh", "-c", "trap 'exit 0' TERM; echo TCP-SNAPSHOT; sleep 30"},
+		"name": "tcp-survivor",
+	})
+	sendFrame(t, tcpClient, protocol.FramePaletteCommand, protocol.PaletteCommand{
+		Kind: "spawn_command",
+		Data: data,
+	})
+	expectFrame(t, tcpClient, protocol.FramePaneSnapshot)
+
+	unixClient := dialDaemon(t, socket)
+	defer unixClient.Close()
+	sendFrame(t, unixClient, protocol.FrameAttach, protocol.Attach{
+		ProjectPath: projectDir,
+		TermSize:    protocol.Size{Cols: 80, Rows: 24},
+	})
+	expectFrame(t, unixClient, protocol.FrameHello)
+
+	cancel()
+	select {
+	case err := <-errCh:
+		if err != nil {
+			t.Fatalf("daemon returned error: %v", err)
+		}
+	case <-time.After(3 * time.Second):
+		t.Fatalf("daemon did not stop")
+	}
+}
+
+func TestDaemonPaneDisplayOwnerSizing(t *testing.T) {
+	t.Setenv("XDG_DATA_HOME", t.TempDir())
+	t.Setenv("XDG_CONFIG_HOME", t.TempDir())
+	projectDir := t.TempDir()
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+
+	reg := newProjectRegistry(preset.Set{}, defaultSettings(), nil, 80, 24)
+	defer reg.Shutdown()
+	project, err := reg.Open(ctx, projectDir)
+	if err != nil {
+		t.Fatalf("open project: %v", err)
+	}
+
+	client1, daemon1 := protocol.NewLoopbackPair()
+	go handleDaemonConn(ctx, cancel, reg, daemon1, "")
+	sendFrame(t, client1, protocol.FrameAttach, protocol.Attach{
+		ProjectPath: projectDir,
+		TermSize:    protocol.Size{Cols: 80, Rows: 24},
+	})
+	expectFrame(t, client1, protocol.FrameHello)
+	expectFrame(t, client1, protocol.FrameProjectList)
+	expectFrame(t, client1, protocol.FrameChrome)
+
+	data, _ := json.Marshal(map[string]any{
+		"argv": []string{"sh", "-c", "trap 'exit 0' TERM; while :; do sleep 1; done"},
+		"name": "owner-pane",
+	})
+	sendFrame(t, client1, protocol.FramePaletteCommand, protocol.PaletteCommand{
+		Kind: "spawn_command",
+		Data: data,
+	})
+	paneID := waitForLifecycleID(t, client1, protocol.LifecycleSpawned, 3*time.Second)
+	snap1 := waitForSnapshot(t, client1, paneID, 3*time.Second)
+	if !snap1.DisplayOwner || snap1.Size != (protocol.Size{Cols: 80, Rows: 24}) {
+		t.Fatalf("owner snapshot = owner:%v size:%+v, want owner true size 80x24", snap1.DisplayOwner, snap1.Size)
+	}
+	waitForEmulatorSize(t, project, paneID, 80, 24)
+
+	client2, daemon2 := protocol.NewLoopbackPair()
+	go handleDaemonConn(ctx, cancel, reg, daemon2, "")
+	sendFrame(t, client2, protocol.FrameAttach, protocol.Attach{
+		ProjectPath: projectDir,
+		TermSize:    protocol.Size{Cols: 100, Rows: 30},
+	})
+	expectFrame(t, client2, protocol.FrameHello)
+	expectFrame(t, client2, protocol.FrameProjectList)
+	expectFrame(t, client2, protocol.FrameChrome)
+	snap2 := waitForSnapshot(t, client2, paneID, 3*time.Second)
+	if snap2.DisplayOwner || snap2.Size != (protocol.Size{Cols: 80, Rows: 24}) {
+		t.Fatalf("viewer snapshot = owner:%v size:%+v, want owner false size 80x24", snap2.DisplayOwner, snap2.Size)
+	}
+	sendFrame(t, client2, protocol.FrameResize, protocol.Resize{Size: protocol.Size{Cols: 100, Rows: 30}})
+	time.Sleep(100 * time.Millisecond)
+	waitForEmulatorSize(t, project, paneID, 80, 24)
+
+	sendFrame(t, client1, protocol.FrameDetach, protocol.Detach{})
+	_ = client1.Close()
+	time.Sleep(100 * time.Millisecond)
+	sendFrame(t, client2, protocol.FrameFocus, protocol.Focus{PaneID: paneID})
+	snap3 := waitForSnapshot(t, client2, paneID, 3*time.Second)
+	if !snap3.DisplayOwner || snap3.Size != (protocol.Size{Cols: 100, Rows: 30}) {
+		t.Fatalf("claimed snapshot = owner:%v size:%+v, want owner true size 100x30", snap3.DisplayOwner, snap3.Size)
+	}
+	waitForEmulatorSize(t, project, paneID, 100, 30)
+
+	sendFrame(t, client2, protocol.FrameDetach, protocol.Detach{})
+	_ = client2.Close()
+}
+
+func waitForSocket(t *testing.T, socket string, errCh <-chan error) {
+	t.Helper()
+	deadline := time.Now().Add(3 * time.Second)
+	for time.Now().Before(deadline) {
+		if _, err := os.Stat(socket); err == nil {
+			return
+		}
+		select {
+		case err := <-errCh:
+			if err != nil && strings.Contains(err.Error(), "operation not permitted") {
+				t.Skipf("unix sockets unavailable in this sandbox: %v", err)
+			}
+			t.Fatalf("daemon exited before creating socket: %v", err)
+		default:
+		}
+		time.Sleep(25 * time.Millisecond)
+	}
+	t.Fatalf("socket %s was not created", socket)
+}
+
+func dialDaemon(t *testing.T, socket string) protocol.Transport {
+	t.Helper()
+	conn, err := net.Dial("unix", socket)
+	if err != nil {
+		t.Fatalf("dial daemon: %v", err)
+	}
+	return protocol.NewConnTransport(conn)
+}
+
+func dialTCPDaemon(t *testing.T, addr string) protocol.Transport {
+	t.Helper()
+	conn, err := net.Dial("tcp", addr)
+	if err != nil {
+		t.Fatalf("dial tcp daemon: %v", err)
+	}
+	return protocol.NewConnTransport(conn)
+}
+
+func waitForTCPAddr(t *testing.T, ready <-chan string, errCh <-chan error) string {
+	t.Helper()
+	select {
+	case addr := <-ready:
+		return addr
+	case err := <-errCh:
+		if err != nil && strings.Contains(err.Error(), "operation not permitted") {
+			t.Skipf("tcp sockets unavailable in this sandbox: %v", err)
+		}
+		t.Fatalf("daemon exited before TCP listener was ready: %v", err)
+	case <-time.After(3 * time.Second):
+		t.Fatalf("tcp listener was not ready")
+	}
+	return ""
+}
+
+func assertTCPAttachDenied(t *testing.T, addr, token string) {
+	t.Helper()
+	client := dialTCPDaemon(t, addr)
+	defer client.Close()
+	sendFrame(t, client, protocol.FrameAttach, protocol.Attach{
+		Token:    token,
+		TermSize: protocol.Size{Cols: 80, Rows: 24},
+	})
+	f := expectFrame(t, client, protocol.FrameError)
+	msg, err := protocol.Decode[protocol.Error](f)
+	if err != nil {
+		t.Fatalf("decode error frame: %v", err)
+	}
+	if !strings.Contains(msg.Message, "auth denied") {
+		t.Fatalf("error message = %q, want auth denied", msg.Message)
+	}
+}
+
+func sendFrame[T any](t *testing.T, tr protocol.Transport, typ protocol.FrameType, payload T) {
+	t.Helper()
+	f, err := protocol.NewFrame(typ, payload)
+	if err != nil {
+		t.Fatalf("frame %s: %v", typ, err)
+	}
+	if err := tr.Send(f); err != nil {
+		t.Fatalf("send %s: %v", typ, err)
+	}
+}
+
+func expectFrame(t *testing.T, tr protocol.Transport, typ protocol.FrameType) protocol.Frame {
+	t.Helper()
+	deadline := time.Now().Add(3 * time.Second)
+	for time.Now().Before(deadline) {
+		f, err, ok := recvFrameWithin(tr, time.Until(deadline))
+		if !ok {
+			break
+		}
+		if err != nil {
+			t.Fatalf("recv %s: %v", typ, err)
+		}
+		if f.Type == typ {
+			return f
+		}
+	}
+	t.Fatalf("frame %s not received", typ)
+	return protocol.Frame{}
+}
+
+func expectChrome(t *testing.T, tr protocol.Transport) protocol.Chrome {
+	t.Helper()
+	f := expectFrame(t, tr, protocol.FrameChrome)
+	chrome, err := protocol.Decode[protocol.Chrome](f)
+	if err != nil {
+		t.Fatalf("decode chrome: %v", err)
+	}
+	return chrome
+}
+
+func waitForLifecycle(t *testing.T, tr protocol.Transport, kind protocol.LifecycleKind, timeout time.Duration) {
+	t.Helper()
+	deadline := time.Now().Add(timeout)
+	for time.Now().Before(deadline) {
+		f, err, ok := recvFrameWithin(tr, time.Until(deadline))
+		if !ok {
+			break
+		}
+		if err != nil {
+			t.Fatalf("recv lifecycle: %v", err)
+		}
+		if f.Type != protocol.FrameLifecycle {
+			continue
+		}
+		msg, err := protocol.Decode[protocol.Lifecycle](f)
+		if err != nil {
+			t.Fatalf("decode lifecycle: %v", err)
+		}
+		if msg.Kind == kind {
+			return
+		}
+	}
+	t.Fatalf("lifecycle %s not received", kind)
+}
+
+func waitForLifecycleID(t *testing.T, tr protocol.Transport, kind protocol.LifecycleKind, timeout time.Duration) string {
+	t.Helper()
+	deadline := time.Now().Add(timeout)
+	for time.Now().Before(deadline) {
+		f, err, ok := recvFrameWithin(tr, time.Until(deadline))
+		if !ok {
+			break
+		}
+		if err != nil {
+			t.Fatalf("recv lifecycle: %v", err)
+		}
+		if f.Type != protocol.FrameLifecycle {
+			continue
+		}
+		msg, err := protocol.Decode[protocol.Lifecycle](f)
+		if err != nil {
+			t.Fatalf("decode lifecycle: %v", err)
+		}
+		if msg.Kind == kind {
+			return msg.ChildID
+		}
+	}
+	t.Fatalf("lifecycle %s not received", kind)
+	return ""
+}
+
+func waitForSnapshot(t *testing.T, tr protocol.Transport, paneID string, timeout time.Duration) protocol.PaneSnapshot {
+	t.Helper()
+	deadline := time.Now().Add(timeout)
+	for time.Now().Before(deadline) {
+		f, err, ok := recvFrameWithin(tr, time.Until(deadline))
+		if !ok {
+			break
+		}
+		if err != nil {
+			t.Fatalf("recv snapshot: %v", err)
+		}
+		if f.Type != protocol.FramePaneSnapshot {
+			continue
+		}
+		msg, err := protocol.Decode[protocol.PaneSnapshot](f)
+		if err != nil {
+			t.Fatalf("decode snapshot: %v", err)
+		}
+		if msg.PaneID == paneID {
+			return msg
+		}
+	}
+	t.Fatalf("snapshot for %s not received", paneID)
+	return protocol.PaneSnapshot{}
+}
+
+func waitForEmulatorSize(t *testing.T, project *Project, paneID string, cols, rows uint16) {
+	t.Helper()
+	deadline := time.Now().Add(3 * time.Second)
+	for time.Now().Before(deadline) {
+		if c := project.Session.FindChild(paneID); c != nil {
+			if em := c.Emulator(); em != nil {
+				gotCols, gotRows := em.Size()
+				if gotCols == cols && gotRows == rows {
+					return
+				}
+			}
+		}
+		time.Sleep(25 * time.Millisecond)
+	}
+	if c := project.Session.FindChild(paneID); c != nil {
+		if em := c.Emulator(); em != nil {
+			gotCols, gotRows := em.Size()
+			t.Fatalf("emulator size = %dx%d, want %dx%d", gotCols, gotRows, cols, rows)
+		}
+	}
+	t.Fatalf("pane %s missing emulator", paneID)
+}
+
+func recvFrameWithin(tr protocol.Transport, timeout time.Duration) (protocol.Frame, error, bool) {
+	type result struct {
+		f   protocol.Frame
+		err error
+	}
+	ch := make(chan result, 1)
+	go func() {
+		f, err := tr.Recv()
+		ch <- result{f: f, err: err}
+	}()
+	select {
+	case r := <-ch:
+		return r.f, r.err, true
+	case <-time.After(timeout):
+		return protocol.Frame{}, nil, false
+	}
+}
+
+func chromeHasProcess(chrome protocol.Chrome, name string) bool {
+	var model struct {
+		Processes []childModel `json:"processes"`
+	}
+	if err := json.Unmarshal(chrome.Model, &model); err != nil {
+		return false
+	}
+	for _, p := range model.Processes {
+		if p.Name == name {
+			return true
+		}
+	}
+	return false
+}

internal/app/debug.go

@@ -0,0 +1,162 @@
+package app
+
+import (
+	"encoding/json"
+	"fmt"
+	"os"
+	"path/filepath"
+	"sync"
+	"time"
+)
+
+// debugCapture implements ChildEventListener and writes structured
+// debug artefacts under a single directory:
+//
+//   - patterm.log    — the existing logf() stream
+//   - events.jsonl   — one JSON object per lifecycle event
+//   - <id>.raw       — raw PTY bytes for each child, by id+name
+//
+// The capture is installed only when --debug=<dir> is set, so default
+// runs pay nothing.
+type debugCapture struct {
+	dir     string
+	logPath string
+
+	mu      sync.Mutex
+	events  *os.File
+	rawByID map[string]*os.File
+}
+
+func openDebugCapture(dir string) (*debugCapture, error) {
+	if err := os.MkdirAll(dir, 0o700); err != nil {
+		return nil, err
+	}
+	logPath := filepath.Join(dir, "patterm.log")
+	// Truncate-style fresh log per run is friendlier for grep'ing one
+	// session. The existing logf opens O_APPEND though, so concurrent
+	// runs against the same dir would interleave — that's on the user.
+	if f, err := os.Create(logPath); err != nil {
+		return nil, err
+	} else {
+		_ = f.Close()
+	}
+	ev, err := os.Create(filepath.Join(dir, "events.jsonl"))
+	if err != nil {
+		return nil, err
+	}
+	dc := &debugCapture{
+		dir:     dir,
+		logPath: logPath,
+		events:  ev,
+		rawByID: make(map[string]*os.File),
+	}
+	dc.writeEvent("session_start", map[string]any{
+		"time": time.Now().Format(time.RFC3339Nano),
+		"pid":  os.Getpid(),
+	})
+	return dc, nil
+}
+
+func (d *debugCapture) LogPath() string { return d.logPath }
+
+func (d *debugCapture) Close() error {
+	d.mu.Lock()
+	defer d.mu.Unlock()
+	d.writeEventLocked("session_end", map[string]any{
+		"time": time.Now().Format(time.RFC3339Nano),
+	})
+	for _, f := range d.rawByID {
+		_ = f.Close()
+	}
+	d.rawByID = nil
+	if d.events != nil {
+		_ = d.events.Close()
+		d.events = nil
+	}
+	return nil
+}
+
+func (d *debugCapture) OnChildSpawned(c *Child) {
+	d.writeEvent("child_spawned", map[string]any{
+		"time":      time.Now().Format(time.RFC3339Nano),
+		"id":        c.ID,
+		"name":      c.Name,
+		"kind":      string(c.Kind),
+		"parent_id": c.ParentID,
+		"preset":    c.PresetRef,
+		"argv":      c.Argv,
+	})
+}
+
+func (d *debugCapture) OnChildExited(c *Child) {
+	d.writeEvent("child_exited", map[string]any{
+		"time":      time.Now().Format(time.RFC3339Nano),
+		"id":        c.ID,
+		"name":      c.Name,
+		"exit_code": c.ExitCode(),
+	})
+	d.mu.Lock()
+	defer d.mu.Unlock()
+	if f, ok := d.rawByID[c.ID]; ok {
+		_ = f.Close()
+		delete(d.rawByID, c.ID)
+	}
+}
+
+func (d *debugCapture) OnChildStateChanged(id string, state IdleState) {
+	d.writeEvent("child_state", map[string]any{
+		"time":  time.Now().Format(time.RFC3339Nano),
+		"id":    id,
+		"state": string(state),
+	})
+}
+
+func (d *debugCapture) OnChildClosed(id string) {
+	d.writeEvent("child_closed", map[string]any{
+		"time": time.Now().Format(time.RFC3339Nano),
+		"id":   id,
+	})
+}
+
+func (d *debugCapture) OnPTYOut(childID string, chunk []byte) {
+	if len(chunk) == 0 {
+		return
+	}
+	d.mu.Lock()
+	defer d.mu.Unlock()
+	f, ok := d.rawByID[childID]
+	if !ok {
+		path := filepath.Join(d.dir, childID+".raw")
+		nf, err := os.Create(path)
+		if err != nil {
+			return
+		}
+		f = nf
+		d.rawByID[childID] = nf
+	}
+	// Listener contract: don't retain chunk past return. Writing now
+	// is fine; the slice's backing buffer is reused for the next read
+	// only after this listener chain completes.
+	_, _ = f.Write(chunk)
+}
+
+func (d *debugCapture) writeEvent(kind string, fields map[string]any) {
+	d.mu.Lock()
+	defer d.mu.Unlock()
+	d.writeEventLocked(kind, fields)
+}
+
+func (d *debugCapture) writeEventLocked(kind string, fields map[string]any) {
+	if d.events == nil {
+		return
+	}
+	if fields == nil {
+		fields = map[string]any{}
+	}
+	fields["event"] = kind
+	enc, err := json.Marshal(fields)
+	if err != nil {
+		return
+	}
+	_, _ = fmt.Fprintln(d.events, string(enc))
+}

internal/app/host.go

@@ -7,6 +7,7 @@ import (
 	"sync"
 	"syscall"
 	"time"
+	"unicode"
 
 	"github.com/hjbdev/patterm/internal/mcp"
 	"github.com/hjbdev/patterm/internal/preset"
@@ -61,12 +62,11 @@ type toolHost struct {
 	prompter  trustPrompter
 	scratch   scratchpadSink
 
-	timersMu  sync.Mutex
-	nextTimer int
+	timers *timerManager
 }
 
 func newToolHost(sess *Session, pads *scratchpad.Store, launcher *Launcher, presets preset.Set, tr *trust.Store, cols, rows uint16) *toolHost {
-	return &toolHost{
+	h := &toolHost{
 		sess:       sess,
 		pads:       pads,
 		launcher:   launcher,
@@ -76,6 +76,31 @@ func newToolHost(sess *Session, pads *scratchpad.Store, launcher *Launcher, pres
 		defaultRow: rows,
 		startedAt:  make(map[string]time.Time),
 	}
+	h.timers = newTimerManager(sess)
+	// Plug the timer manager into the session's state-change fan-out so
+	// idle-aware timers fire when watched children transition into idle.
+	// Tests can construct a host with a nil session for sizing checks —
+	// those never run timers, so the subscribe is skipped.
+	if sess != nil {
+		sess.Subscribe(timerListenerAdapter{m: h.timers})
+	}
+	return h
+}
+
+// timerListenerAdapter forwards OnChildStateChanged and OnChildClosed
+// into the timer manager and ignores the other ChildEventListener
+// methods. The session's listener API is by-interface, so we wrap
+// the manager rather than make it implement the full surface.
+type timerListenerAdapter struct{ m *timerManager }
+
+func (a timerListenerAdapter) OnChildSpawned(*Child)   {}
+func (a timerListenerAdapter) OnChildExited(*Child)    {}
+func (a timerListenerAdapter) OnPTYOut(string, []byte) {}
+func (a timerListenerAdapter) OnChildStateChanged(id string, st IdleState) {
+	a.m.onChildStateChanged(id, st)
+}
+func (a timerListenerAdapter) OnChildClosed(id string) {
+	a.m.onChildClosed(id)
 }
 
 func (h *toolHost) SetSize(cols, rows uint16) {
@@ -374,11 +399,11 @@ func (h *toolHost) GetProcessOutput(callerID, processID, mode string, sinceOffse
 		if c.Kind == KindAgent {
 			txt = applyChromeTrim(txt, h.chromeHintsFor(c.PresetRef))
 		}
-		out.Content = txt
+		out.Content = normalizeGridText(txt)
 		return out, nil
 	case "stream":
 		b, end := c.StreamRead(sinceOffset)
-		out.Content = stripANSI(string(b))
+		out.Content = string(stripANSIBytes(nil, b))
 		out.NewOffset = end
 		return out, nil
 	default:
@@ -409,10 +434,10 @@ func (h *toolHost) SearchOutput(callerID, processID, pattern, kind string, limit
 		return mcp.SearchResult{}, mcp.Errorf(mcp.ErrorKindInvalidArgs, "regex: %v", err)
 	}
 	b, _ := c.StreamRead(0)
-	text := string(b)
 	if kind == "rendered" {
-		text = stripANSI(text)
+		b = stripANSIBytes(nil, b)
 	}
+	text := string(b)
 	lines := strings.Split(text, "\n")
 	matches := make([]mcp.SearchMatch, 0, limit)
 	truncated := false
@@ -440,10 +465,19 @@ func (h *toolHost) WaitForPattern(callerID, processID, pattern string, timeoutSe
 	if scope == "" {
 		scope = "grid"
 	}
+	if scope != "grid" && scope != "scrollback" {
+		return false, "", mcp.Errorf(mcp.ErrorKindInvalidArgs, "unknown scope %q (want grid|scrollback)", scope)
+	}
 	deadline := time.Now().Add(time.Duration(timeoutSeconds * float64(time.Second)))
-	tick := time.NewTicker(50 * time.Millisecond)
-	defer tick.Stop()
-	for {
+
+	// chunkWake fires on every PTY chunk for the target child. The
+	// fallback timer guarantees we still re-check on grid-only sweeps
+	// where the cursor position changed without a fresh chunk landing.
+	wake := newChunkNotifier(c.ID)
+	h.sess.Subscribe(wake)
+	defer h.sess.Unsubscribe(wake)
+
+	check := func() (bool, string) {
 		text := ""
 		switch scope {
 		case "grid":
@@ -454,23 +488,77 @@ func (h *toolHost) WaitForPattern(callerID, processID, pattern string, timeoutSe
 			}
 		case "scrollback":
 			b, _ := c.StreamRead(0)
-			text = stripANSI(string(b))
-		default:
-			return false, "", mcp.Errorf(mcp.ErrorKindInvalidArgs, "unknown scope %q (want grid|scrollback)", scope)
+			text = string(stripANSIBytes(nil, b))
 		}
 		if m := re.FindString(text); m != "" {
-			return true, m, nil
+			return true, m
 		}
-		if time.Now().After(deadline) {
+		return false, ""
+	}
+
+	if ok, m := check(); ok {
+		return true, m, nil
+	}
+	for {
+		remaining := time.Until(deadline)
+		if remaining <= 0 {
 			return false, "", nil
 		}
-		<-tick.C
+		// Long fallback tick — the chunk notifier wakes us promptly
+		// on fresh PTY output; the timer is only there for cases
+		// where grid state shifted without a new chunk.
+		wait := 500 * time.Millisecond
+		if remaining < wait {
+			wait = remaining
+		}
+		select {
+		case <-wake.fired:
+		case <-time.After(wait):
+		}
+		if ok, m := check(); ok {
+			return true, m, nil
+		}
 		if !c.IsLive() && c.Status() != StatusStopped {
 			return false, "", nil
 		}
 	}
 }
 
+// chunkNotifier is a one-shot-per-chunk wake channel listener.
+// Registers via Session.Subscribe; emits a non-blocking signal on
+// `fired` for every PTY chunk emitted by the target child. Used by
+// WaitForPattern to avoid 50ms-tick polling of the entire ring/grid.
+type chunkNotifier struct {
+	childID string
+	fired   chan struct{}
+}
+
+func newChunkNotifier(childID string) *chunkNotifier {
+	return &chunkNotifier{childID: childID, fired: make(chan struct{}, 1)}
+}
+
+func (n *chunkNotifier) OnChildSpawned(*Child) {}
+func (n *chunkNotifier) OnChildExited(c *Child) {
+	if c.ID != n.childID {
+		return
+	}
+	select {
+	case n.fired <- struct{}{}:
+	default:
+	}
+}
+func (n *chunkNotifier) OnPTYOut(id string, chunk []byte) {
+	if id != n.childID {
+		return
+	}
+	select {
+	case n.fired <- struct{}{}:
+	default:
+	}
+}
+func (n *chunkNotifier) OnChildStateChanged(string, IdleState) {}
+func (n *chunkNotifier) OnChildClosed(string)                  {}
+
 func (h *toolHost) GetProcessPorts(callerID, processID string) ([]mcp.PortSighting, error) {
 	c := h.sess.FindChild(processID)
 	if c == nil {
@@ -664,40 +752,72 @@ func (h *toolHost) RequestHumanAttention(callerID, processID, reason string) err
 	return nil
 }
 
+// TimerWait is the legacy fire-and-forget delay timer. It now wraps
+// TimerSet with an empty body — defaultFireFn substitutes the
+// "[system] Your timer […] has completed." line so behaviour matches
+// the original API. New callers should use timer_set with an explicit
+// body.
 func (h *toolHost) TimerWait(callerID string, seconds float64, label string) (string, error) {
-	caller := h.sess.FindChild(callerID)
-	if caller == nil {
-		return "", mcp.Errorf(mcp.ErrorKindNotFound, "caller %q not known to patterm", callerID)
+	return h.timers.TimerSet(callerID, "", label, seconds)
+}
+
+func (h *toolHost) TimerSet(callerID string, args mcp.TimerSetArgs) (mcp.TimerHandle, error) {
+	owner := resolveTimerOwner(callerID, args.OwnerProcessID)
+	id, err := h.timers.TimerSet(owner, args.Body, args.Label, args.Seconds)
+	if err != nil {
+		return mcp.TimerHandle{}, err
 	}
-	h.timersMu.Lock()
-	h.nextTimer++
-	id := fmt.Sprintf("t%d", h.nextTimer)
-	h.timersMu.Unlock()
-	if label == "" {
-		label = id
+	return mcp.TimerHandle{ID: id}, nil
+}
+
+func (h *toolHost) TimerFireWhenIdleAny(callerID string, args mcp.TimerFireWhenIdleArgs) (mcp.TimerFireWhenIdleResponse, error) {
+	owner := resolveTimerOwner(callerID, args.OwnerProcessID)
+	return h.timers.TimerFireWhenIdleAny(owner, args.Body, args.Label, args.Watched, args.MaxWaitSeconds)
+}
+
+func (h *toolHost) TimerFireWhenIdleAll(callerID string, args mcp.TimerFireWhenIdleArgs) (mcp.TimerFireWhenIdleResponse, error) {
+	owner := resolveTimerOwner(callerID, args.OwnerProcessID)
+	return h.timers.TimerFireWhenIdleAll(owner, args.Body, args.Label, args.Watched, args.MaxWaitSeconds)
+}
+
+// resolveTimerOwner picks the owner process for a timer. Explicit
+// owner_process_id wins; otherwise the caller's own id is used.
+// Top-level MCP clients (no callerID) must provide owner_process_id
+// explicitly.
+func resolveTimerOwner(callerID, explicit string) string {
+	if explicit != "" {
+		return explicit
 	}
-	go func() {
-		time.Sleep(time.Duration(seconds * float64(time.Second)))
-		if !caller.IsLive() {
-			return
-		}
-		line := fmt.Sprintf("[system] Your timer [%s] has completed.\r", label)
-		_ = caller.InjectAsOrchestrator([]byte(line))
-	}()
-	return id, nil
+	return callerID
+}
+
+func (h *toolHost) TimerCancel(callerID, id string) error {
+	return h.timers.TimerCancel(callerID, id)
+}
+
+func (h *toolHost) TimerPause(callerID, id string) error {
+	return h.timers.TimerPause(callerID, id)
+}
+
+func (h *toolHost) TimerResume(callerID, id string) error {
+	return h.timers.TimerResume(callerID, id)
+}
+
+func (h *toolHost) TimerList(callerID string) ([]mcp.TimerInfo, error) {
+	return h.timers.TimerList(callerID), nil
 }
 
 // ───────────────────────────────────────────────────────────────────
 // Scratchpads / Meta
 // ───────────────────────────────────────────────────────────────────
 
-func (h *toolHost) ScratchpadList() ([]scratchpad.Entry, error) { return h.pads.List() }
+func (h *toolHost) ScratchpadList(string) ([]scratchpad.Entry, error) { return h.pads.List() }
 
-func (h *toolHost) ScratchpadRead(name string) (string, string, error) {
+func (h *toolHost) ScratchpadRead(_ string, name string) (string, string, error) {
 	return h.pads.Read(name)
 }
 
-func (h *toolHost) ScratchpadWrite(name, content, expectedRevision string) (string, error) {
+func (h *toolHost) ScratchpadWrite(_, name, content, expectedRevision string) (string, error) {
 	rev, err := h.pads.Write(name, content, expectedRevision)
 	if err == nil && h.scratch != nil {
 		h.scratch.scratchpadsChanged()
@@ -705,7 +825,7 @@ func (h *toolHost) ScratchpadWrite(name, content, expectedRevision string) (stri
 	return rev, err
 }
 
-func (h *toolHost) ScratchpadAppend(name, content string) error {
+func (h *toolHost) ScratchpadAppend(_, name, content string) error {
 	err := h.pads.Append(name, content)
 	if err == nil && h.scratch != nil {
 		h.scratch.scratchpadsChanged()
@@ -713,6 +833,14 @@ func (h *toolHost) ScratchpadAppend(name, content string) error {
 	return err
 }
 
+func (h *toolHost) ScratchpadDelete(_, name string) error {
+	err := h.pads.Delete(name)
+	if err == nil && h.scratch != nil {
+		h.scratch.scratchpadsChanged()
+	}
+	return err
+}
+
 func (h *toolHost) WhoAmI(callerID string) mcp.WhoAmI {
 	w := mcp.WhoAmI{
 		ProcessID: callerID,
@@ -755,6 +883,10 @@ func (h *toolHost) processInfoOf(c *Child) mcp.ProcessInfo {
 		t := h.trust.IsTrusted(c.PresetRef)
 		info.Trusted = &t
 	}
+	if s := c.IdleState(); s != StateUnknown {
+		info.IdleState = string(s)
+		info.IdleReason = c.IdleReason()
+	}
 	return info
 }
 
@@ -887,6 +1019,98 @@ func stripANSI(s string) string {
 	return ansiRegexp.ReplaceAllString(s, "")
 }
 
+func normalizeGridText(s string) string {
+	s = strings.ReplaceAll(s, "\r\n", "\n")
+	s = strings.ReplaceAll(s, "\r", "\n")
+
+	lines := strings.Split(s, "\n")
+	out := make([]string, 0, len(lines))
+	pendingBlank := false
+	for _, line := range lines {
+		line = strings.TrimRightFunc(line, unicode.IsSpace)
+		if line == "" {
+			if len(out) > 0 {
+				pendingBlank = true
+			}
+			continue
+		}
+		if pendingBlank {
+			out = append(out, "")
+			pendingBlank = false
+		}
+		out = append(out, line)
+	}
+	return strings.Join(out, "\n")
+}
+
+// stripANSIBytes is the byte-slice form of stripANSI. Skips the
+// string conversion and the regex DFA — useful when the caller will
+// itself walk the result line-by-line (SearchOutput) or feed it to a
+// pattern match (WaitForPattern scrollback). Recognises the same
+// shapes the regex did:
+//   - `\x1b[ <params> <intermediate> <final-byte>` (CSI / SGR)
+//   - `\x1b<final-byte>` for `@..._` (one-byte escapes)
+//   - `\x07` (BEL)
+//
+// The dst slice is reused if cap is sufficient; the returned slice
+// is what callers should use.
+func stripANSIBytes(dst, src []byte) []byte {
+	if cap(dst) < len(src) {
+		dst = make([]byte, 0, len(src))
+	} else {
+		dst = dst[:0]
+	}
+	for i := 0; i < len(src); {
+		b := src[i]
+		if b == 0x07 {
+			i++
+			continue
+		}
+		if b != 0x1b {
+			dst = append(dst, b)
+			i++
+			continue
+		}
+		// ESC-led sequence.
+		if i+1 >= len(src) {
+			// Stranded ESC at end of buffer — drop it.
+			i++
+			continue
+		}
+		next := src[i+1]
+		if next != '[' {
+			// One-byte ESC sequence (`\x1b<final>` where final is
+			// `@..._` per the regex; we drop anything that follows).
+			if next >= 0x40 && next <= 0x5f {
+				i += 2
+				continue
+			}
+			// Anything else after ESC: drop the ESC, keep walking.
+			i++
+			continue
+		}
+		// CSI: parameters [0x30..0x3f]*, intermediate [0x20..0x2f]*,
+		// final [0x40..0x7e].
+		j := i + 2
+		for j < len(src) && src[j] >= 0x30 && src[j] <= 0x3f {
+			j++
+		}
+		for j < len(src) && src[j] >= 0x20 && src[j] <= 0x2f {
+			j++
+		}
+		if j < len(src) && src[j] >= 0x40 && src[j] <= 0x7e {
+			i = j + 1
+			continue
+		}
+		// Incomplete CSI — the regex form falls back to its
+		// `\x1b<final>` rule and matches `\x1b[` (`[` is 0x5b, inside
+		// 0x40..0x5f), consuming the two-byte prefix and leaving the
+		// pending params/intermediate bytes intact. Match that.
+		i += 2
+	}
+	return dst
+}
+
 // availableToolsForRole — SPEC §7 whoami exposes the list a caller can
 // invoke from its current role. Sub-agents lose `spawn_agent` (§8
 // two-level-tree rule).
@@ -897,8 +1121,10 @@ func availableToolsForRole(role mcp.CallerRole) []string {
 		"list_processes", "get_process_status", "get_project_status",
 		"get_process_output", "get_process_raw_output", "search_output",
 		"wait_for_pattern", "get_process_ports",
-		"send_input", "send_message", "request_human_attention", "timer_wait",
-		"scratchpad_list", "scratchpad_read", "scratchpad_write", "scratchpad_append",
+		"send_input", "send_message", "request_human_attention",
+		"timer_wait", "timer_set", "timer_fire_when_idle_any", "timer_fire_when_idle_all",
+		"timer_cancel", "timer_pause", "timer_resume", "timer_list",
+		"scratchpad_list", "scratchpad_read", "scratchpad_write", "scratchpad_append", "scratchpad_delete",
 		"whoami", "help",
 	}
 	if role == mcp.RoleOrchestrator {
@@ -922,13 +1148,13 @@ func helpFor(topic string) mcp.HelpResponse {
 	case "spawning":
 		return mcp.HelpResponse{
 			Topic:        "spawning",
-			Content:      "spawn_agent launches another vendor LLM CLI as a sub-agent (orchestrator only). spawn_process(kind: command, preset: …) starts a stored command; spawn_process(kind: terminal) opens a shell. Command presets need trust the first time — you'll get needs_trust until the human accepts. Whatever you spawn is yours to clean up — see help('lifecycle').",
+			Content:      "spawn_agent launches another vendor LLM CLI as a sub-agent (orchestrator only). spawn_process(kind: command, preset: …) starts a stored command; spawn_process(kind: terminal) opens a shell. Command presets need trust the first time — you'll get needs_trust until the human accepts. ANTI-PATTERNS: do not shell out to `claude` / `codex` / `opencode` (or any other agent CLI) yourself, and do not pipe JSON-RPC into patterm's Unix socket via perl / nc / socat / curl. Either path bypasses caller-identity and the new agent reads back as a stray top-level tab instead of your child — call spawn_agent through the MCP transport you were initialised on. Whatever you spawn is yours to clean up — see help('lifecycle').",
 			RelatedTools: []string{"spawn_agent", "spawn_process", "start_process", "restart_process", "close_process"},
 		}
 	case "lifecycle":
 		return mcp.HelpResponse{
-			Topic: "lifecycle",
-			Content: "You own the processes you spawn. When a sub-agent has finished its task (it reports back via send_message, or you've collected what you need from it) call close_process on its process_id to remove the entry and tear down the PTY. Same goes for spawn_process children: command/terminal panes you started are not auto-reclaimed when their work completes. close_process is the normal cleanup path; stop_process(signal) is for sending a signal without removing the entry; start_process re-attaches an exited command preset. Leaving idle sub-agents around wastes vendor tokens and clutters the host — close them as soon as you're done. Sub-agents themselves are reminded (via the [system: …] preface on their first prompt) to clean up anything they created before reporting done.",
+			Topic:        "lifecycle",
+			Content:      "You own the processes you spawn. When a sub-agent has finished its task (it reports back via send_message, or you've collected what you need from it) call close_process on its process_id to remove the entry and tear down the PTY. Same goes for spawn_process children: command/terminal panes you started are not auto-reclaimed when their work completes. close_process is the normal cleanup path; stop_process(signal) is for sending a signal without removing the entry; start_process re-attaches an exited command preset. Leaving idle sub-agents around wastes vendor tokens and clutters the host — close them as soon as you're done. Sub-agents themselves are reminded (via the [system: …] preface on their first prompt) to clean up anything they created before reporting done.",
 			RelatedTools: []string{"close_process", "stop_process", "start_process", "list_processes", "get_process_status"},
 		}
 	case "inspection":
@@ -945,27 +1171,42 @@ func helpFor(topic string) mcp.HelpResponse {
 		}
 	case "coordination":
 		return mcp.HelpResponse{
-			Topic:        "coordination",
-			Content:      "send_message tags the message with the caller's role (parent → [orchestrator], child → [sub-agent:<name>]). Siblings must route through their parent. request_human_attention raises a UI notification when you can't safely decide.",
-			RelatedTools: []string{"send_message", "request_human_attention"},
+			Topic: "coordination",
+			Content: "send_message tags the message with the caller's role (parent → [orchestrator], child → [sub-agent:<name>]). Siblings must route through their parent. request_human_attention raises a UI notification when you can't safely decide.\n\n" +
+				"Reply routing: a sub-agent's reply to your send_message lands in YOUR pane tagged `[sub-agent:<name>]`, not in the sub-agent's output. Anti-pattern: `wait_for_pattern(sub_agent, …)` to wait for a reply — the sub-agent is already idle, its output won't change, and the call spins to timeout. Pattern: send_message → timer_fire_when_idle_any([sub_agent_id], body=\"[system] sub-agent finished\") → when the timer fires, the reply is already queued as your next user turn (or visible via get_process_output on your own pane).",
+			RelatedTools: []string{"send_message", "request_human_attention", "timer_fire_when_idle_any", "timer_fire_when_idle_all"},
 		}
 	case "scratchpads":
 		return mcp.HelpResponse{
 			Topic:        "scratchpads",
-			Content:      "Project-scoped markdown files. Read returns content + revision; pass that back as expected_revision on write to get last-write-wins-with-detection. Append is unconditional.",
-			RelatedTools: []string{"scratchpad_list", "scratchpad_read", "scratchpad_write", "scratchpad_append"},
+			Content:      "Project-scoped markdown files. Read returns content + revision; pass that back as expected_revision on write to get last-write-wins-with-detection. Append is unconditional; delete removes a pad by name.",
+			RelatedTools: []string{"scratchpad_list", "scratchpad_read", "scratchpad_write", "scratchpad_append", "scratchpad_delete"},
 		}
 	case "timers":
 		return mcp.HelpResponse{
-			Topic:        "timers",
-			Content:      "timer_wait returns a timer_id immediately and injects `[system] Your timer [<label>] has completed.` into your pane when it fires. Use it instead of sleeping in your own process.",
-			RelatedTools: []string{"timer_wait"},
+			Topic: "timers",
+			Content: "Timers fire by injecting your chosen body (or a default `[system] Your timer […] has completed.` line) back into your pane as a fresh user turn. Use them instead of sleeping in your own process. " +
+				"timer_wait / timer_set schedule a delay timer (timer_set lets you set body+label). " +
+				"timer_fire_when_idle_any fires when any watched process becomes idle (already-idle watchers are excluded from the baseline). " +
+				"timer_fire_when_idle_all fires when every watched process is idle; if all are idle at registration the response is already_satisfied with no pending timer. " +
+				"timer_cancel / timer_pause / timer_resume manage outstanding timers; resume re-checks idle conditions in case a watcher went idle while paused. " +
+				"timer_list shows your pending and paused timers.",
+			RelatedTools: []string{
+				"timer_wait", "timer_set",
+				"timer_fire_when_idle_any", "timer_fire_when_idle_all",
+				"timer_cancel", "timer_pause", "timer_resume", "timer_list",
+			},
 		}
 	case "readiness":
 		return mcp.HelpResponse{
-			Topic:        "readiness",
-			Content:      "A pane is 'idle' once nothing has been written to its PTY for ~1s (SPEC §11). Treat idle as a signal to read, not a guarantee of completion. wait_for_pattern lets you wait on a known terminal marker for stronger evidence.",
-			RelatedTools: []string{"wait_for_pattern", "get_process_status"},
+			Topic: "readiness",
+			Content: "A pane is 'idle' once nothing has been written to its PTY for ~1s (SPEC §11). Treat idle as a signal to read, not a guarantee of completion.\n\n" +
+				"Waiting for a sub-agent's reply (canonical pattern):\n" +
+				"  1. send_message(sub_agent_id, request)\n" +
+				"  2. timer_fire_when_idle_any(watched=[sub_agent_id], body=\"[system] sub-agent done\")\n" +
+				"  3. When the timer fires you re-enter as a fresh user turn; the sub-agent's reply is already in your own pane tagged `[sub-agent:<name>]` (read via get_process_output on yourself if you need it explicitly).\n\n" +
+				"wait_for_pattern is for waiting on text a process emits in its OWN output (a shell prompt, a build's \"tests passed\" line). It does NOT see send_message replies, because those land in the caller's pane, not the target's — calling wait_for_pattern on a sub-agent to wait for its reply deadlocks until timeout.",
+			RelatedTools: []string{"wait_for_pattern", "get_process_status", "timer_fire_when_idle_any", "send_message"},
 		}
 	case "permissions":
 		return mcp.HelpResponse{

internal/app/host_test.go

@@ -3,6 +3,8 @@ package app
 import (
 	"strings"
 	"testing"
+
+	"github.com/hjbdev/patterm/internal/mcp"
 )
 
 // mkChild builds a Child without starting a PTY. Use sparingly — the
@@ -164,6 +166,47 @@ func TestHelpSpawningPointsAtLifecycle(t *testing.T) {
 	}
 }
 
+// TestAvailableToolsAdvertisesAllTimerTools makes sure orchestrators
+// and sub-agents discover the full timer surface via whoami — not just
+// timer_wait. Otherwise agents using whoami for orientation would never
+// learn about timer_set, timer_fire_when_idle_*, timer_pause/resume,
+// timer_cancel, and timer_list.
+func TestAvailableToolsAdvertisesAllTimerTools(t *testing.T) {
+	want := []string{
+		"timer_wait", "timer_set",
+		"timer_fire_when_idle_any", "timer_fire_when_idle_all",
+		"timer_cancel", "timer_pause", "timer_resume", "timer_list",
+	}
+	for _, role := range []mcp.CallerRole{mcp.RoleOrchestrator, mcp.RoleSubAgent} {
+		tools := availableToolsForRole(role)
+		for _, w := range want {
+			if !containsString(tools, w) {
+				t.Fatalf("role %q missing %q in available tools: %v", role, w, tools)
+			}
+		}
+	}
+}
+
+// TestHelpTimersDocumentsAllTools mirrors the whoami check for the
+// help("timers") topic — the related-tools list must enumerate every
+// timer_* tool so callers reading help can dispatch them.
+func TestHelpTimersDocumentsAllTools(t *testing.T) {
+	resp := helpFor("timers")
+	if resp.Topic != "timers" {
+		t.Fatalf("topic: %q", resp.Topic)
+	}
+	want := []string{
+		"timer_wait", "timer_set",
+		"timer_fire_when_idle_any", "timer_fire_when_idle_all",
+		"timer_cancel", "timer_pause", "timer_resume", "timer_list",
+	}
+	for _, w := range want {
+		if !containsString(resp.RelatedTools, w) {
+			t.Fatalf("timers help missing %q in related tools: %v", w, resp.RelatedTools)
+		}
+	}
+}
+
 func containsString(haystack []string, needle string) bool {
 	for _, s := range haystack {
 		if s == needle {
@@ -172,4 +215,3 @@ func containsString(haystack []string, needle string) bool {
 	}
 	return false
 }
-

internal/app/idle.go

@@ -0,0 +1,228 @@
+package app
+
+import (
+	"regexp"
+
+	"github.com/hjbdev/patterm/internal/preset"
+)
+
+// IdleState is the classifier's opinion about what a child is doing.
+// Inspired by Solo's five-state model. ERROR is a terminal state — set
+// when a child exits non-zero or matches an error-promoter regex —
+// while the other four reflect transient runtime state.
+type IdleState string
+
+const (
+	StateUnknown    IdleState = ""
+	StateIdle       IdleState = "idle"
+	StateWorking    IdleState = "working"
+	StateThinking   IdleState = "thinking"
+	StatePermission IdleState = "permission"
+	StateError      IdleState = "error"
+)
+
+// IdleStrategy picks the primary signal used to decide idle vs working.
+// Promoter regexes can override this on top.
+type IdleStrategy string
+
+const (
+	StrategyOutputActivity    IdleStrategy = "output_activity"
+	StrategyOSCTitleStability IdleStrategy = "osc_title_stability"
+	StrategyOSCTitleStatus    IdleStrategy = "osc_title_status"
+)
+
+// defaultIdleThresholdMS is used when a preset doesn't override it.
+const defaultIdleThresholdMS = 2000
+
+// resolvedIdleDetection is the compiled, runtime-ready form of a
+// preset.IdleDetection block. Built once at child spawn and held
+// read-only by the classifier; regex patterns are compiled here so the
+// hot path doesn't pay for it.
+type resolvedIdleDetection struct {
+	strategy        IdleStrategy
+	idleThresholdMS int64
+
+	titleStatusMap map[string]IdleState
+
+	permissionRegexes []*regexp.Regexp
+	thinkingRegexes   []*regexp.Regexp
+	errorRegexes      []*regexp.Regexp
+}
+
+// resolveIdleDetection compiles a preset.IdleDetection (which may be
+// nil) into the runtime form. Unknown strategies fall back to
+// output_activity. Pattern compile errors are skipped silently — the
+// preset loader is responsible for surfacing them as warnings.
+func resolveIdleDetection(cfg *preset.IdleDetection) *resolvedIdleDetection {
+	r := &resolvedIdleDetection{
+		strategy:        StrategyOutputActivity,
+		idleThresholdMS: defaultIdleThresholdMS,
+	}
+	if cfg == nil {
+		return r
+	}
+	switch IdleStrategy(cfg.Strategy) {
+	case StrategyOSCTitleStability, StrategyOSCTitleStatus, StrategyOutputActivity:
+		r.strategy = IdleStrategy(cfg.Strategy)
+	}
+	if cfg.IdleThresholdMS > 0 {
+		r.idleThresholdMS = int64(cfg.IdleThresholdMS)
+	}
+	if len(cfg.TitleStatusMap) > 0 {
+		r.titleStatusMap = make(map[string]IdleState, len(cfg.TitleStatusMap))
+		for k, v := range cfg.TitleStatusMap {
+			switch IdleState(v) {
+			case StateIdle, StateWorking, StateThinking, StatePermission, StateError:
+				r.titleStatusMap[k] = IdleState(v)
+			}
+		}
+	}
+	r.permissionRegexes = compilePatterns(cfg.PermissionPatterns)
+	r.thinkingRegexes = compilePatterns(cfg.ThinkingPatterns)
+	r.errorRegexes = compilePatterns(cfg.ErrorPatterns)
+	return r
+}
+
+func compilePatterns(ps []string) []*regexp.Regexp {
+	if len(ps) == 0 {
+		return nil
+	}
+	out := make([]*regexp.Regexp, 0, len(ps))
+	for _, p := range ps {
+		if p == "" {
+			continue
+		}
+		re, err := regexp.Compile(p)
+		if err != nil {
+			continue
+		}
+		out = append(out, re)
+	}
+	return out
+}
+
+// classify computes the IdleState from the inputs the classifier loop
+// has already gathered. Pure function so it's easy to unit-test.
+//
+// Resolution order:
+//  1. terminal: process exited non-zero → error (latched)
+//  2. error-promoter regex match in recent output → error
+//  3. permission-promoter regex match → permission
+//  4. thinking-promoter regex match → thinking
+//  5. strategy-specific base classification (idle vs working).
+//
+// inputs:
+//   - exited: whether the child process has exited
+//   - exitNonZero: whether the exit was non-zero (only meaningful when exited)
+//   - idleMS: ms since the last PTY output
+//   - titleIdleMS: ms since the last OSC title change (0 if no title yet)
+//   - title: current OSC title
+//   - tail: recent output bytes for regex matching
+//   - screen: current rendered screen text for persistent prompt matching
+func classify(cfg *resolvedIdleDetection, exited, exitNonZero bool, idleMS, titleIdleMS int64, title string, tail, screen []byte) (IdleState, string) {
+	if exited {
+		if exitNonZero {
+			return StateError, "process exited non-zero"
+		}
+		return StateIdle, "process exited cleanly"
+	}
+	if cfg == nil {
+		cfg = &resolvedIdleDetection{strategy: StrategyOutputActivity, idleThresholdMS: defaultIdleThresholdMS}
+	}
+	if len(tail) > 0 || len(screen) > 0 {
+		if matchAny(cfg.errorRegexes, tail, screen) {
+			return StateError, "error regex matched"
+		}
+		if matchAny(cfg.permissionRegexes, tail, screen) {
+			return StatePermission, "permission regex matched"
+		}
+		if matchAny(cfg.thinkingRegexes, tail, screen) {
+			return StateThinking, "thinking regex matched"
+		}
+	}
+	threshold := cfg.idleThresholdMS
+	switch cfg.strategy {
+	case StrategyOSCTitleStatus:
+		// First try the title-status map; if no match, fall back to
+		// title-stability behaviour so we still produce idle/working.
+		if s, ok := matchTitleStatus(cfg.titleStatusMap, title); ok {
+			return s, "title status match"
+		}
+		fallthrough
+	case StrategyOSCTitleStability:
+		// If we've never seen a title, fall back to output activity so
+		// we don't latch in idle while the child is clearly running.
+		if titleIdleMS == 0 {
+			return baseStateFromIdleMS(idleMS, threshold)
+		}
+		return baseStateFromIdleMS(titleIdleMS, threshold)
+	default: // output_activity
+		return baseStateFromIdleMS(idleMS, threshold)
+	}
+}
+
+func baseStateFromIdleMS(idleMS, threshold int64) (IdleState, string) {
+	// idleMS == 0 means "no writes yet" (per Child.IdleMS) — treat as
+	// not-idle so we don't classify a freshly-spawned child as idle.
+	if idleMS == 0 {
+		return StateWorking, "no activity yet"
+	}
+	if idleMS < threshold {
+		return StateWorking, "recent activity"
+	}
+	return StateIdle, "quiet for threshold"
+}
+
+func matchAny(res []*regexp.Regexp, texts ...[]byte) bool {
+	for _, re := range res {
+		for _, text := range texts {
+			if len(text) > 0 && re.Match(text) {
+				return true
+			}
+		}
+	}
+	return false
+}
+
+func matchTitleStatus(m map[string]IdleState, title string) (IdleState, bool) {
+	if len(m) == 0 || title == "" {
+		return StateUnknown, false
+	}
+	for k, v := range m {
+		if k == "" {
+			continue
+		}
+		if containsFold(title, k) {
+			return v, true
+		}
+	}
+	return StateUnknown, false
+}
+
+// containsFold reports whether s contains sub, case-insensitively.
+// Cheap implementation suitable for short titles.
+func containsFold(s, sub string) bool {
+	if len(sub) == 0 {
+		return true
+	}
+	if len(sub) > len(s) {
+		return false
+	}
+	ls, lsub := lower(s), lower(sub)
+	for i := 0; i+len(lsub) <= len(ls); i++ {
+		if ls[i:i+len(lsub)] == lsub {
+			return true
+		}
+	}
+	return false
+}
+
+func lower(s string) string {
+	b := []byte(s)
+	for i, c := range b {
+		if c >= 'A' && c <= 'Z' {
+			b[i] = c + 32
+		}
+	}
+	return string(b)
+}

internal/app/idle_test.go

@@ -0,0 +1,132 @@
+package app
+
+import (
+	"regexp"
+	"testing"
+)
+
+func mustCompile(t *testing.T, p string) *regexp.Regexp {
+	t.Helper()
+	re, err := regexp.Compile(p)
+	if err != nil {
+		t.Fatalf("regex %q: %v", p, err)
+	}
+	return re
+}
+
+func TestClassifyOutputActivity(t *testing.T) {
+	cfg := &resolvedIdleDetection{strategy: StrategyOutputActivity, idleThresholdMS: 2000}
+
+	cases := []struct {
+		name   string
+		idleMS int64
+		want   IdleState
+	}{
+		{"fresh-spawn no writes", 0, StateWorking},
+		{"recent activity", 500, StateWorking},
+		{"under threshold", 1999, StateWorking},
+		{"at threshold", 2000, StateIdle},
+		{"over threshold", 5000, StateIdle},
+	}
+	for _, tc := range cases {
+		t.Run(tc.name, func(t *testing.T) {
+			got, _ := classify(cfg, false, false, tc.idleMS, 0, "", nil, nil)
+			if got != tc.want {
+				t.Fatalf("got %q want %q", got, tc.want)
+			}
+		})
+	}
+}
+
+func TestClassifyTitleStability(t *testing.T) {
+	cfg := &resolvedIdleDetection{strategy: StrategyOSCTitleStability, idleThresholdMS: 2000}
+	// Title change recent → working.
+	if got, _ := classify(cfg, false, false, 9999, 500, "step 3", nil, nil); got != StateWorking {
+		t.Fatalf("recent title change: got %q", got)
+	}
+	// Title stable past threshold → idle.
+	if got, _ := classify(cfg, false, false, 9999, 5000, "step 3", nil, nil); got != StateIdle {
+		t.Fatalf("stable title: got %q", got)
+	}
+	// No title yet: fall back to output activity.
+	if got, _ := classify(cfg, false, false, 100, 0, "", nil, nil); got != StateWorking {
+		t.Fatalf("no title yet, recent output: got %q", got)
+	}
+	if got, _ := classify(cfg, false, false, 5000, 0, "", nil, nil); got != StateIdle {
+		t.Fatalf("no title yet, output idle: got %q", got)
+	}
+}
+
+func TestClassifyTitleStabilityThinkingPatternOverridesIdle(t *testing.T) {
+	cfg := &resolvedIdleDetection{
+		strategy:        StrategyOSCTitleStability,
+		idleThresholdMS: 2000,
+		thinkingRegexes: []*regexp.Regexp{mustCompile(t, `(?i)esc to interrupt`)},
+	}
+	screen := []byte("• Working (5s • esc to interrupt)")
+	if got, _ := classify(cfg, false, false, 9999, 5000, "codex", nil, screen); got != StateThinking {
+		t.Fatalf("thinking screen marker: got %q want %q", got, StateThinking)
+	}
+	if got, _ := classify(cfg, false, false, 9999, 5000, "codex", nil, []byte(">_")); got != StateIdle {
+		t.Fatalf("stable title without marker: got %q want %q", got, StateIdle)
+	}
+}
+
+func TestClassifyTitleStatus(t *testing.T) {
+	cfg := &resolvedIdleDetection{
+		strategy:        StrategyOSCTitleStatus,
+		idleThresholdMS: 2000,
+		titleStatusMap: map[string]IdleState{
+			"thinking":   StateThinking,
+			"permission": StatePermission,
+			"error":      StateError,
+		},
+	}
+	if got, _ := classify(cfg, false, false, 9999, 500, "Thinking…", nil, nil); got != StateThinking {
+		t.Fatalf("thinking title: got %q", got)
+	}
+	if got, _ := classify(cfg, false, false, 9999, 500, "Waiting for permission", nil, nil); got != StatePermission {
+		t.Fatalf("permission title: got %q", got)
+	}
+	// No match in map → fall back to stability.
+	if got, _ := classify(cfg, false, false, 9999, 5000, "ready", nil, nil); got != StateIdle {
+		t.Fatalf("unmatched title, stable: got %q", got)
+	}
+}
+
+func TestClassifyPromoterRegex(t *testing.T) {
+	cfg := &resolvedIdleDetection{
+		strategy:          StrategyOutputActivity,
+		idleThresholdMS:   2000,
+		permissionRegexes: []*regexp.Regexp{mustCompile(t, `Approve\?`)},
+		errorRegexes:      []*regexp.Regexp{mustCompile(t, `panic:`)},
+		thinkingRegexes:   []*regexp.Regexp{mustCompile(t, `Thinking`)},
+	}
+	// Permission promoter beats idle.
+	if got, _ := classify(cfg, false, false, 5000, 0, "", []byte("Approve? [y/n]"), nil); got != StatePermission {
+		t.Fatalf("permission promoter: got %q", got)
+	}
+	// Error trumps permission.
+	if got, _ := classify(cfg, false, false, 5000, 0, "", []byte("panic: bad\nApprove?"), nil); got != StateError {
+		t.Fatalf("error promoter beats permission: got %q", got)
+	}
+	// Thinking promoter on idle output.
+	if got, _ := classify(cfg, false, false, 5000, 0, "", []byte("Thinking…"), nil); got != StateThinking {
+		t.Fatalf("thinking promoter: got %q", got)
+	}
+	// Rendered-screen prompts still promote even when the raw tail no
+	// longer contains the original prompt bytes.
+	if got, _ := classify(cfg, false, false, 100, 0, "", []byte("Calling patterm..."), []byte("Approve? [y/n]")); got != StatePermission {
+		t.Fatalf("screen permission promoter: got %q", got)
+	}
+}
+
+func TestClassifyExitTerminal(t *testing.T) {
+	cfg := &resolvedIdleDetection{strategy: StrategyOutputActivity, idleThresholdMS: 2000}
+	if got, _ := classify(cfg, true, true, 0, 0, "", nil, nil); got != StateError {
+		t.Fatalf("non-zero exit: got %q", got)
+	}
+	if got, _ := classify(cfg, true, false, 0, 0, "", nil, nil); got != StateIdle {
+		t.Fatalf("clean exit: got %q", got)
+	}
+}

internal/app/keymatch.go

@@ -40,6 +40,36 @@ type csiuKey struct {
 	event int
 }
 
+// parseSGRMouseWheel decodes the parameter run from an SGR-encoded
+// mouse press (`CSI < button ; col ; row M`) and returns a row delta
+// when the event is a scroll wheel. Wheel-up returns -wheelStep,
+// wheel-down returns +wheelStep. Modifier bits in the button code
+// (shift=4, alt=8, ctrl=16) are stripped before matching, so e.g.
+// shift+wheel still scrolls. Non-wheel buttons return false.
+func parseSGRMouseWheel(params []byte) (int, bool) {
+	const wheelStep = 3
+	// Button code runs up to the first ';'.
+	end := 0
+	for end < len(params) && params[end] != ';' {
+		end++
+	}
+	if end == 0 {
+		return 0, false
+	}
+	btn, err := strconv.Atoi(string(params[:end]))
+	if err != nil {
+		return 0, false
+	}
+	if btn&64 == 0 {
+		return 0, false
+	}
+	// Bit 0 selects up (0) vs down (1) for wheel events.
+	if btn&1 == 0 {
+		return -wheelStep, true
+	}
+	return wheelStep, true
+}
+
 // decodeCSIu parses the parameter string of a `CSI ... u` sequence.
 // The kitty shape is:
 //

internal/app/keymatch_test.go

@@ -39,6 +39,32 @@ func TestMatchCtrlK(t *testing.T) {
 	}
 }
 
+func TestParseSGRMouseWheel(t *testing.T) {
+	cases := []struct {
+		params string
+		want   int
+		ok     bool
+	}{
+		{"64;1;1", -3, true},  // wheel up
+		{"65;1;1", 3, true},   // wheel down
+		{"68;1;1", -3, true},  // shift+wheel up
+		{"69;1;1", 3, true},   // shift+wheel down
+		{"80;1;1", -3, true},  // ctrl+wheel up
+		{"81;1;1", 3, true},   // ctrl+wheel down
+		{"0;5;7", 0, false},   // left press
+		{"2;5;7", 0, false},   // right press
+		{"32;5;7", 0, false},  // drag
+		{"", 0, false},        // empty
+		{"abc;1;1", 0, false}, // garbage button
+	}
+	for _, c := range cases {
+		got, ok := parseSGRMouseWheel([]byte(c.params))
+		if ok != c.ok || got != c.want {
+			t.Errorf("parseSGRMouseWheel(%q) = (%d,%v), want (%d,%v)", c.params, got, ok, c.want, c.ok)
+		}
+	}
+}
+
 func TestMatchCtrlKConsecutive(t *testing.T) {
 	// Two kitty Ctrl-K sequences back to back, the chord case.
 	chunk := []byte("\x1b[107;5u\x1b[107;5u")

internal/app/launch.go

@@ -9,6 +9,7 @@ import (
 	"sync"
 	"time"
 
+	"github.com/hjbdev/patterm/internal/persist"
 	"github.com/hjbdev/patterm/internal/preset"
 )
 
@@ -126,14 +127,15 @@ func (l *Launcher) LaunchAgent(p *preset.Preset, displayName, initialPrompt, par
 	// Spawn with the chosen identity.
 	cols, rows := l.size()
 	c, err := l.sess.Spawn(SpawnSpec{
-		Kind:         KindAgent,
-		Argv:         argv,
-		Env:          env,
-		Name:         displayName,
-		ParentID:     parentID,
-		PresetRef:    p.Name,
-		Identity:     identity,
-		CleanupPaths: cleanupPaths,
+		Kind:          KindAgent,
+		Argv:          argv,
+		Env:           env,
+		Name:          displayName,
+		ParentID:      parentID,
+		PresetRef:     p.Name,
+		Identity:      identity,
+		CleanupPaths:  cleanupPaths,
+		IdleDetection: resolveIdleDetection(p.IdleDetection),
 	}, cols, rows)
 	if err != nil {
 		cleanup()
@@ -170,15 +172,20 @@ func (l *Launcher) LaunchCommandPreset(p *preset.Preset, displayName, parentID s
 		env = append(env, k+"="+v)
 	}
 	cols, rows := l.size()
-	return l.sess.Spawn(SpawnSpec{
-		Kind:      KindCommand,
-		Argv:      p.ResolvedArgv(),
-		Env:       env,
-		Name:      displayName,
-		ParentID:  parentID,
-		WorkDir:   p.WorkingDir,
-		PresetRef: p.Name,
+	c, err := l.sess.Spawn(SpawnSpec{
+		Kind:          KindCommand,
+		Argv:          p.ResolvedArgv(),
+		Env:           env,
+		Name:          displayName,
+		ParentID:      parentID,
+		WorkDir:       p.WorkingDir,
+		PresetRef:     p.Name,
+		IdleDetection: resolveIdleDetection(p.IdleDetection),
 	}, cols, rows)
+	if err != nil {
+		return nil, err
+	}
+	return c, nil
 }
 
 // LaunchCommandArgv spawns a freeform-argv command entry. Trust gating
@@ -202,6 +209,33 @@ func (l *Launcher) LaunchCommandArgv(argv []string, displayName, parentID, workD
 	}, cols, rows)
 }
 
+// RestoreCommand re-spawns a persisted top-level command entry. If
+// the entry has a PresetRef and the preset still exists, the spawn
+// goes through LaunchCommandPreset (so preset.Env / WorkingDir stay
+// authoritative). Otherwise the saved argv runs directly via
+// LaunchCommandArgv with shell=false — entries that were originally
+// `shell: true` were already wrapped into `["sh","-lc",...]` before
+// persistence, so re-wrapping isn't needed.
+//
+// Returns the freshly minted Child. The caller is responsible for
+// setting auto-restart back on the returned entry.
+func (l *Launcher) RestoreCommand(e persist.Entry, presets preset.Set) (*Child, error) {
+	if e.PresetRef != "" {
+		for _, p := range presets.Processes {
+			if p.Name == e.PresetRef {
+				return l.LaunchCommandPreset(p, e.Name, "")
+			}
+		}
+		// Preset has been deleted since the entry was saved. Fall
+		// through to argv-based restore using whatever the saved
+		// command looked like at the time.
+	}
+	if len(e.Argv) == 0 {
+		return nil, fmt.Errorf("restore: entry %s has no argv", e.ID)
+	}
+	return l.LaunchCommandArgv(e.Argv, e.Name, "", e.WorkDir, nil, false)
+}
+
 // LaunchTerminal spawns a bare interactive shell. SPEC §7 kind=terminal.
 // argv defaults to $SHELL -i when empty.
 func (l *Launcher) LaunchTerminal(argv []string, displayName, parentID, workDir string, env []string) (*Child, error) {
@@ -227,15 +261,11 @@ func (l *Launcher) LaunchTerminal(argv []string, displayName, parentID, workDir
 }
 
 func (l *Launcher) writeMCPConfig(identity string) (string, error) {
-	dir, err := preset.ConfigDir()
+	dir, err := mcpRuntimeDir(identity)
 	if err != nil {
 		return "", err
 	}
-	dir = filepath.Join(dir, "mcp")
-	if err := os.MkdirAll(dir, 0o700); err != nil {
-		return "", err
-	}
-	path := filepath.Join(dir, identity+".json")
+	path := filepath.Join(dir, "mcp.json")
 	cfg := map[string]any{
 		"mcpServers": map[string]any{
 			"patterm": map[string]any{

internal/app/launch_test.go

@@ -0,0 +1,30 @@
+package app
+
+import (
+	"os"
+	"path/filepath"
+	"strings"
+	"testing"
+)
+
+func TestWriteMCPConfigUsesRuntimeDir(t *testing.T) {
+	runtimeDir := t.TempDir()
+	configHome := filepath.Join(t.TempDir(), "config")
+	t.Setenv("XDG_RUNTIME_DIR", runtimeDir)
+	t.Setenv("XDG_CONFIG_HOME", configHome)
+
+	l := &Launcher{bin: "patterm", mcpSocket: "/tmp/patterm.sock"}
+	path, err := l.writeMCPConfig("abc123")
+	if err != nil {
+		t.Fatalf("writeMCPConfig: %v", err)
+	}
+	if !strings.HasPrefix(path, filepath.Join(runtimeDir, "patterm", "agents", "abc123")) {
+		t.Fatalf("path = %q, want under runtime dir", path)
+	}
+	if _, err := os.Stat(path); err != nil {
+		t.Fatalf("config file stat: %v", err)
+	}
+	if _, err := os.Stat(filepath.Join(configHome, "patterm")); !os.IsNotExist(err) {
+		t.Fatalf("writeMCPConfig created XDG config dir or unexpected stat error: %v", err)
+	}
+}

internal/app/layout_test.go

@@ -14,10 +14,10 @@ func TestTerminalLayoutWideUsesMainViewport(t *testing.T) {
 	if l.childCols() != 91 {
 		t.Fatalf("child cols: got %d want 91", l.childCols())
 	}
-	if l.childRows() != 37 {
-		t.Fatalf("child rows: got %d want 37", l.childRows())
+	if l.childRows() != 36 {
+		t.Fatalf("child rows: got %d want 36", l.childRows())
 	}
-	if l.mainTop != 3 || l.statusRow != 40 {
+	if l.mainTop != 4 || l.statusRow != 40 {
 		t.Fatalf("unexpected vertical chrome: mainTop=%d statusRow=%d", l.mainTop, l.statusRow)
 	}
 }
@@ -30,8 +30,8 @@ func TestTerminalLayoutNarrowHidesSidebar(t *testing.T) {
 	if l.childCols() != 38 {
 		t.Fatalf("child cols: got %d want 38", l.childCols())
 	}
-	if l.childRows() != 9 {
-		t.Fatalf("child rows: got %d want 9", l.childRows())
+	if l.childRows() != 8 {
+		t.Fatalf("child rows: got %d want 8", l.childRows())
 	}
 }
 
@@ -46,13 +46,13 @@ func TestSpawnSizingUsesViewportDimensions(t *testing.T) {
 	l := newTerminalLayout(120, 40)
 	launcher := NewLauncher(nil, "", l.childCols(), l.childRows())
 	cols, rows := launcher.size()
-	if cols != 91 || rows != 37 {
-		t.Fatalf("launcher size: got %dx%d want 91x37", cols, rows)
+	if cols != 91 || rows != 36 {
+		t.Fatalf("launcher size: got %dx%d want 91x36", cols, rows)
 	}
 
 	host := newToolHost(nil, nil, nil, preset.Set{}, nil, l.childCols(), l.childRows())
 	cols, rows = host.size()
-	if cols != 91 || rows != 37 {
-		t.Fatalf("tool host size: got %dx%d want 91x37", cols, rows)
+	if cols != 91 || rows != 36 {
+		t.Fatalf("tool host size: got %dx%d want 91x36", cols, rows)
 	}
 }

internal/app/markdown.go

@@ -0,0 +1,483 @@
+package app
+
+import (
+	"strings"
+	"unicode"
+	"unicode/utf8"
+)
+
+// renderMarkdownLines turns a scratchpad's text into a slice of
+// terminal rows, each at most `cols` visible columns wide and ready to
+// paint (style codes included, trailing reset where needed, no
+// newline). The renderer covers the markdown subset most likely to
+// appear in scratchpad notes: headings (#, ##, ###), bold (**x**),
+// inline code (`x`), fenced code blocks (```), bullet/numbered lists,
+// blockquotes (> ), horizontal rules, and links rendered as their
+// text. Plain text passes through unchanged.
+func renderMarkdownLines(content string, cols int) []string {
+	if cols < 1 {
+		cols = 1
+	}
+	var out []string
+	inFence := false
+	for _, raw := range strings.Split(content, "\n") {
+		line := strings.TrimRight(raw, "\r")
+		trimmed := strings.TrimSpace(line)
+
+		if strings.HasPrefix(trimmed, "```") {
+			inFence = !inFence
+			out = append(out, mdFenceRule(cols))
+			continue
+		}
+		if inFence {
+			out = append(out, mdCodeBlockLines(line, cols)...)
+			continue
+		}
+		if trimmed == "" {
+			out = append(out, "")
+			continue
+		}
+		if isMDHRule(trimmed) {
+			out = append(out, styleBorder+strings.Repeat("─", cols)+styleReset)
+			continue
+		}
+		if body, level := parseMDHeading(line); level > 0 {
+			style := mdHeadingStyle(level)
+			out = append(out, wrapInline(parseInline(body), style, cols)...)
+			continue
+		}
+		if body, ok := parseBlockquote(line); ok {
+			prefix := styleAccent + "│ " + styleReset
+			lines := wrapInline(parseInline(body), styleHint, cols-2)
+			if len(lines) == 0 {
+				out = append(out, prefix)
+				continue
+			}
+			for _, l := range lines {
+				out = append(out, prefix+l)
+			}
+			continue
+		}
+		if marker, body, ok := parseListItem(line); ok {
+			prefix := mdBulletPrefix(marker)
+			indent := strings.Repeat(" ", mdVisibleLen(prefix))
+			lines := wrapInline(parseInline(body), "", cols-mdVisibleLen(prefix))
+			if len(lines) == 0 {
+				out = append(out, prefix)
+				continue
+			}
+			for i, l := range lines {
+				if i == 0 {
+					out = append(out, prefix+l)
+				} else {
+					out = append(out, indent+l)
+				}
+			}
+			continue
+		}
+		out = append(out, wrapInline(parseInline(line), "", cols)...)
+	}
+	return out
+}
+
+func mdHeadingStyle(level int) string {
+	switch level {
+	case 1:
+		return styleActive + styleBold
+	case 2:
+		return styleBold + styleAccent
+	default:
+		return styleBold
+	}
+}
+
+func mdBulletPrefix(marker string) string {
+	if isOrderedMarker(marker) {
+		return styleAccent + marker + " " + styleReset
+	}
+	return styleAccent + "• " + styleReset
+}
+
+func mdFenceRule(cols int) string {
+	if cols < 2 {
+		return styleBorder + strings.Repeat("─", cols) + styleReset
+	}
+	return styleBorder + strings.Repeat("─", cols) + styleReset
+}
+
+// mdCodeBlockLines emits one rendered row per (wrapped) source line
+// inside a fenced code block, prefixed with a thin accent gutter so the
+// block reads as one visual unit.
+func mdCodeBlockLines(line string, cols int) []string {
+	gutter := styleAccent + "│" + styleReset + " "
+	body := line
+	avail := cols - 2
+	if avail < 1 {
+		avail = 1
+	}
+	chunks := wrapPlain(body, avail)
+	if len(chunks) == 0 {
+		return []string{gutter}
+	}
+	out := make([]string, 0, len(chunks))
+	for _, c := range chunks {
+		out = append(out, gutter+"\x1b[38;5;180m"+c+styleReset)
+	}
+	return out
+}
+
+func isMDHRule(s string) bool {
+	if len(s) < 3 {
+		return false
+	}
+	c := s[0]
+	if c != '-' && c != '_' && c != '*' {
+		return false
+	}
+	for i := 0; i < len(s); i++ {
+		if s[i] != c && s[i] != ' ' {
+			return false
+		}
+	}
+	count := 0
+	for i := 0; i < len(s); i++ {
+		if s[i] == c {
+			count++
+		}
+	}
+	return count >= 3
+}
+
+func parseMDHeading(line string) (string, int) {
+	i := 0
+	for i < len(line) && line[i] == ' ' && i < 3 {
+		i++
+	}
+	level := 0
+	for i+level < len(line) && line[i+level] == '#' && level < 6 {
+		level++
+	}
+	if level == 0 {
+		return "", 0
+	}
+	rest := line[i+level:]
+	if rest != "" && rest[0] != ' ' {
+		return "", 0
+	}
+	return strings.TrimSpace(rest), level
+}
+
+func parseBlockquote(line string) (string, bool) {
+	t := strings.TrimLeft(line, " ")
+	if !strings.HasPrefix(t, ">") {
+		return "", false
+	}
+	rest := strings.TrimPrefix(t, ">")
+	rest = strings.TrimPrefix(rest, " ")
+	return rest, true
+}
+
+func parseListItem(line string) (marker, body string, ok bool) {
+	t := strings.TrimLeft(line, " ")
+	if len(t) >= 2 && (t[0] == '-' || t[0] == '*' || t[0] == '+') && t[1] == ' ' {
+		return string(t[0]), t[2:], true
+	}
+	// Ordered: digits then "." then space.
+	j := 0
+	for j < len(t) && t[j] >= '0' && t[j] <= '9' {
+		j++
+	}
+	if j > 0 && j+1 < len(t) && t[j] == '.' && t[j+1] == ' ' {
+		return t[:j+1], t[j+2:], true
+	}
+	return "", "", false
+}
+
+func isOrderedMarker(m string) bool {
+	if len(m) < 2 {
+		return false
+	}
+	if m[len(m)-1] != '.' {
+		return false
+	}
+	for i := 0; i < len(m)-1; i++ {
+		if m[i] < '0' || m[i] > '9' {
+			return false
+		}
+	}
+	return true
+}
+
+// mdSpan is one styled run of plain text. style is an SGR prefix
+// applied at the start; the renderer emits styleReset between adjacent
+// spans of differing style and at end-of-line.
+type mdSpan struct {
+	text  string
+	style string
+}
+
+// parseInline turns one source line into styled spans. Recognises:
+//   - **bold** / __bold__ → bold span
+//   - `code` → inline code span
+//   - [text](url) → text rendered as accent+underline
+//
+// Unmatched delimiters are passed through as literal characters so a
+// stray `*` or backtick doesn't swallow the rest of the line.
+func parseInline(line string) []mdSpan {
+	var spans []mdSpan
+	var buf strings.Builder
+	flush := func(style string) {
+		if buf.Len() == 0 {
+			return
+		}
+		spans = append(spans, mdSpan{text: buf.String(), style: style})
+		buf.Reset()
+	}
+	i := 0
+	for i < len(line) {
+		c := line[i]
+		switch {
+		case c == '`':
+			if end := strings.IndexByte(line[i+1:], '`'); end >= 0 {
+				flush("")
+				spans = append(spans, mdSpan{text: line[i+1 : i+1+end], style: "\x1b[38;5;180m"})
+				i += end + 2
+				continue
+			}
+		case c == '*' && i+1 < len(line) && line[i+1] == '*':
+			if end := strings.Index(line[i+2:], "**"); end >= 0 {
+				flush("")
+				inner := parseInline(line[i+2 : i+2+end])
+				for _, s := range inner {
+					st := s.style
+					if st == "" {
+						st = styleBold
+					}
+					spans = append(spans, mdSpan{text: s.text, style: st})
+				}
+				i += end + 4
+				continue
+			}
+		case c == '_' && i+1 < len(line) && line[i+1] == '_':
+			if end := strings.Index(line[i+2:], "__"); end >= 0 {
+				flush("")
+				inner := parseInline(line[i+2 : i+2+end])
+				for _, s := range inner {
+					st := s.style
+					if st == "" {
+						st = styleBold
+					}
+					spans = append(spans, mdSpan{text: s.text, style: st})
+				}
+				i += end + 4
+				continue
+			}
+		case c == '[':
+			if close := strings.IndexByte(line[i+1:], ']'); close >= 0 {
+				rest := line[i+1+close+1:]
+				if strings.HasPrefix(rest, "(") {
+					if pclose := strings.IndexByte(rest[1:], ')'); pclose >= 0 {
+						flush("")
+						label := line[i+1 : i+1+close]
+						spans = append(spans, mdSpan{text: label, style: styleAccent + "\x1b[4m"})
+						i += 1 + close + 1 + 1 + pclose + 1
+						continue
+					}
+				}
+			}
+		}
+		buf.WriteByte(c)
+		i++
+	}
+	flush("")
+	return spans
+}
+
+// wrapInline lays out styled spans across one or more terminal rows of
+// `cols` visible columns each. Each output row is prefixed with
+// `lineStyle` so the caller can theme an entire wrapped paragraph
+// (headings, blockquotes) with one SGR. Wrapping prefers word
+// boundaries; oversized tokens hard-cut at the column boundary.
+func wrapInline(spans []mdSpan, lineStyle string, cols int) []string {
+	if cols < 1 {
+		cols = 1
+	}
+	var out []string
+	var b strings.Builder
+	written := 0
+	curStyle := ""
+
+	startLine := func() {
+		b.Reset()
+		written = 0
+		curStyle = ""
+		if lineStyle != "" {
+			b.WriteString(lineStyle)
+			curStyle = lineStyle
+		}
+	}
+	finishLine := func() {
+		if b.Len() == 0 && lineStyle == "" {
+			out = append(out, "")
+			return
+		}
+		b.WriteString(styleReset)
+		out = append(out, b.String())
+	}
+
+	startLine()
+	writeChar := func(r rune, st string) {
+		if curStyle != st {
+			b.WriteString(styleReset)
+			if lineStyle != "" {
+				b.WriteString(lineStyle)
+			}
+			if st != "" {
+				b.WriteString(st)
+			}
+			curStyle = st
+		}
+		b.WriteRune(r)
+		written += runeCellWidth(r)
+	}
+
+	for _, sp := range spans {
+		st := sp.style
+		// Tokenize span into words+spaces for word-boundary wrapping.
+		text := sp.text
+		for len(text) > 0 {
+			r, size := utf8.DecodeRuneInString(text)
+			// Take a run of either spaces or non-spaces.
+			isSpace := unicode.IsSpace(r)
+			j := 0
+			w := 0
+			for j < len(text) {
+				rr, sz := utf8.DecodeRuneInString(text[j:])
+				if unicode.IsSpace(rr) != isSpace {
+					break
+				}
+				j += sz
+				w += runeCellWidth(rr)
+			}
+			tok := text[:j]
+			text = text[j:]
+			_ = r
+			_ = size
+
+			if isSpace {
+				if written == 0 {
+					// Drop leading whitespace at line start.
+					continue
+				}
+				if written+w > cols {
+					finishLine()
+					startLine()
+					continue
+				}
+				for _, rr := range tok {
+					writeChar(rr, st)
+				}
+				continue
+			}
+			// Non-space token. If it fits, append; else wrap.
+			if w <= cols {
+				if written+w > cols {
+					// Trim trailing spaces written so far before wrap.
+					finishLine()
+					startLine()
+				}
+				for _, rr := range tok {
+					writeChar(rr, st)
+				}
+				continue
+			}
+			// Token longer than a full row: hard-cut.
+			for _, rr := range tok {
+				cw := runeCellWidth(rr)
+				if written+cw > cols {
+					finishLine()
+					startLine()
+				}
+				writeChar(rr, st)
+			}
+		}
+	}
+	finishLine()
+	if len(out) == 0 {
+		out = append(out, "")
+	}
+	return out
+}
+
+// wrapPlain wraps a literal string (no styling) at a `cols` visible
+// column budget. Used by code-block rendering, which preserves the raw
+// line verbatim.
+func wrapPlain(line string, cols int) []string {
+	if cols < 1 {
+		cols = 1
+	}
+	if line == "" {
+		return []string{""}
+	}
+	var out []string
+	var b strings.Builder
+	written := 0
+	for _, r := range line {
+		w := runeCellWidth(r)
+		if written+w > cols {
+			out = append(out, b.String())
+			b.Reset()
+			written = 0
+		}
+		b.WriteRune(r)
+		written += w
+	}
+	if b.Len() > 0 {
+		out = append(out, b.String())
+	}
+	return out
+}
+
+// runeCellWidth is a tiny approximation of terminal cell width: 0 for
+// non-printables, 1 for the common case. Wide East-Asian and emoji
+// runes would ideally be 2, but pads in practice are Latin/symbol text;
+// landing a precise width walk is left for when we see a real case.
+func runeCellWidth(r rune) int {
+	if r == 0 || r == '\r' || r == '\n' {
+		return 0
+	}
+	if r < 0x20 || r == 0x7f {
+		return 0
+	}
+	return 1
+}
+
+// mdVisibleLen counts visible columns in a string with embedded SGR
+// escapes — the inverse of the writer that produces them.
+func mdVisibleLen(s string) int {
+	n := 0
+	i := 0
+	for i < len(s) {
+		if s[i] == 0x1b {
+			j := i + 1
+			if j < len(s) && s[j] == '[' {
+				j++
+				for j < len(s) && !isCSIFinal(s[j]) {
+					j++
+				}
+				if j < len(s) {
+					j++
+				}
+				i = j
+				continue
+			}
+			i = j
+			continue
+		}
+		r, size := utf8.DecodeRuneInString(s[i:])
+		n += runeCellWidth(r)
+		i += size
+	}
+	return n
+}
+

internal/app/markdown_test.go

@@ -0,0 +1,93 @@
+package app
+
+import (
+	"strings"
+	"testing"
+)
+
+func TestRenderMarkdownLines_Heading(t *testing.T) {
+	lines := renderMarkdownLines("# Hello", 40)
+	if len(lines) != 1 {
+		t.Fatalf("heading should be 1 line, got %d (%v)", len(lines), lines)
+	}
+	if !strings.Contains(lines[0], "Hello") {
+		t.Errorf("heading text missing: %q", lines[0])
+	}
+	if !strings.Contains(lines[0], "\x1b[1m") {
+		t.Errorf("heading not bold: %q", lines[0])
+	}
+}
+
+func TestRenderMarkdownLines_BulletWrapping(t *testing.T) {
+	src := "- alpha beta gamma delta epsilon"
+	lines := renderMarkdownLines(src, 14)
+	if len(lines) < 2 {
+		t.Fatalf("expected wrap into 2+ lines, got %d: %v", len(lines), lines)
+	}
+	if !strings.Contains(lines[0], "•") {
+		t.Errorf("first line should carry bullet, got %q", lines[0])
+	}
+	if strings.Contains(lines[1], "•") {
+		t.Errorf("continuation should not repeat bullet: %q", lines[1])
+	}
+}
+
+func TestRenderMarkdownLines_InlineCode(t *testing.T) {
+	lines := renderMarkdownLines("call `foo()` now", 40)
+	if len(lines) != 1 {
+		t.Fatalf("expected one line, got %d", len(lines))
+	}
+	if !strings.Contains(lines[0], "foo()") {
+		t.Errorf("inline code text missing: %q", lines[0])
+	}
+	if !strings.Contains(lines[0], "\x1b[38;5;180m") {
+		t.Errorf("inline code style missing: %q", lines[0])
+	}
+}
+
+func TestRenderMarkdownLines_FencedCode(t *testing.T) {
+	src := "before\n```\nfn main() {\n}\n```\nafter"
+	lines := renderMarkdownLines(src, 40)
+	// Two fence rules + two code rows + before + after = at least 5 lines.
+	if len(lines) < 5 {
+		t.Fatalf("expected fenced block to produce >=5 rows, got %d: %v", len(lines), lines)
+	}
+	foundCode := false
+	for _, l := range lines {
+		if strings.Contains(l, "fn main()") {
+			foundCode = true
+			break
+		}
+	}
+	if !foundCode {
+		t.Errorf("code block content missing from output: %v", lines)
+	}
+}
+
+func TestRenderMarkdownLines_HardWrap(t *testing.T) {
+	src := strings.Repeat("a", 50)
+	lines := renderMarkdownLines(src, 10)
+	if len(lines) < 5 {
+		t.Fatalf("expected long line to wrap into >=5 rows, got %d: %v", len(lines), lines)
+	}
+}
+
+func TestRenderMarkdownLines_PreservesBlankLines(t *testing.T) {
+	src := "para1\n\npara2"
+	lines := renderMarkdownLines(src, 40)
+	if len(lines) != 3 {
+		t.Fatalf("expected 3 rows, got %d: %v", len(lines), lines)
+	}
+	if lines[1] != "" {
+		t.Errorf("middle row should be empty, got %q", lines[1])
+	}
+}
+
+func TestMDVisibleLen(t *testing.T) {
+	if got := mdVisibleLen("\x1b[1mfoo\x1b[0m"); got != 3 {
+		t.Errorf("mdVisibleLen styled: want 3 got %d", got)
+	}
+	if got := mdVisibleLen("hello"); got != 5 {
+		t.Errorf("mdVisibleLen plain: want 5 got %d", got)
+	}
+}

internal/app/marquee.go

@@ -0,0 +1,123 @@
+package app
+
+import (
+	"sync"
+	"time"
+)
+
+// Phase ordering of the marquee state machine: hold the head, scroll
+// one cell per marqueeStep until the tail is visible, hold the tail,
+// snap back to the head.
+const (
+	phaseHoldStart = iota
+	phaseScroll
+	phaseHoldEnd
+)
+
+const (
+	marqueeHoldStart = time.Second
+	marqueeStep      = 150 * time.Millisecond
+	marqueeHoldEnd   = time.Second
+)
+
+// marqueeState drives the focused sidebar row's pause-scroll-pause
+// animation. State is wall-clock anchored (since), not tick-count
+// anchored, so a missed tick yields a slightly later frame rather
+// than a skipped one.
+type marqueeState struct {
+	mu      sync.Mutex
+	id      string
+	nameLen int
+	budget  int
+	state   int
+	offset  int
+	since   time.Time
+}
+
+// step advances the state machine for the row identified by id with
+// the given visible name length (in runes) and column budget. It
+// returns the current scroll offset, whether the row is animating
+// (i.e. nameLen > budget), and how long until the next visual change.
+//
+// When id changes, or nameLen <= budget, the state machine resets to
+// phaseHoldStart with offset 0 anchored at now.
+func (m *marqueeState) step(id string, nameLen, budget int, now time.Time) (offset int, animating bool, nextWake time.Duration) {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+
+	if id != m.id || nameLen != m.nameLen || budget != m.budget {
+		m.id = id
+		m.nameLen = nameLen
+		m.budget = budget
+		m.state = phaseHoldStart
+		m.offset = 0
+		m.since = now
+	}
+
+	if nameLen <= budget || budget <= 0 {
+		return 0, false, 0
+	}
+
+	maxOffset := nameLen - budget
+
+	for {
+		elapsed := now.Sub(m.since)
+		switch m.state {
+		case phaseHoldStart:
+			if elapsed < marqueeHoldStart {
+				return 0, true, marqueeHoldStart - elapsed
+			}
+			m.state = phaseScroll
+			m.since = m.since.Add(marqueeHoldStart)
+			continue
+		case phaseScroll:
+			steps := int(elapsed / marqueeStep)
+			if steps >= maxOffset {
+				m.offset = maxOffset
+				m.state = phaseHoldEnd
+				m.since = m.since.Add(time.Duration(maxOffset) * marqueeStep)
+				continue
+			}
+			m.offset = steps
+			rem := marqueeStep - (elapsed % marqueeStep)
+			return m.offset, true, rem
+		case phaseHoldEnd:
+			if elapsed < marqueeHoldEnd {
+				return maxOffset, true, marqueeHoldEnd - elapsed
+			}
+			m.state = phaseHoldStart
+			m.offset = 0
+			m.since = m.since.Add(marqueeHoldEnd)
+			continue
+		default:
+			m.state = phaseHoldStart
+			m.offset = 0
+			m.since = now
+			return 0, true, marqueeHoldStart
+		}
+	}
+}
+
+// active reports whether the marquee currently has an overflowing row
+// to animate. The marquee ticker goroutine uses this to gate dirty
+// flag flips so an idle sidebar costs nothing.
+func (m *marqueeState) active() bool {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	return m.id != "" && m.nameLen > m.budget && m.budget > 0
+}
+
+// reset clears all state, forcing the next step() call to start a
+// fresh phaseHoldStart. Call this when focus changes so the newly
+// focused row begins with a full head-hold instead of inheriting
+// whatever phase the previous focus was in.
+func (m *marqueeState) reset() {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	m.id = ""
+	m.nameLen = 0
+	m.budget = 0
+	m.state = phaseHoldStart
+	m.offset = 0
+	m.since = time.Time{}
+}

internal/app/marquee_test.go

@@ -0,0 +1,161 @@
+package app
+
+import (
+	"testing"
+	"time"
+)
+
+func TestMarqueeStepFits(t *testing.T) {
+	var m marqueeState
+	now := time.Unix(0, 0)
+	off, animating, _ := m.step("a", 5, 10, now)
+	if animating {
+		t.Fatalf("expected no animation when name fits in budget")
+	}
+	if off != 0 {
+		t.Fatalf("expected offset 0, got %d", off)
+	}
+}
+
+func TestMarqueePhaseProgression(t *testing.T) {
+	var m marqueeState
+	// name 10 runes, budget 5 → maxOffset = 5.
+	const nameLen, budget = 10, 5
+	t0 := time.Unix(0, 0)
+
+	// At t0: phaseHoldStart, offset 0, animating.
+	off, anim, wake := m.step("row", nameLen, budget, t0)
+	if off != 0 || !anim || wake != marqueeHoldStart {
+		t.Fatalf("t0: off=%d anim=%v wake=%v", off, anim, wake)
+	}
+
+	// Just before hold expires: still offset 0.
+	off, anim, _ = m.step("row", nameLen, budget, t0.Add(marqueeHoldStart-time.Millisecond))
+	if off != 0 || !anim {
+		t.Fatalf("pre-expiry hold: off=%d anim=%v", off, anim)
+	}
+
+	// At hold expiry + 1 step: should have transitioned to scroll, offset 1.
+	off, anim, _ = m.step("row", nameLen, budget, t0.Add(marqueeHoldStart+marqueeStep))
+	if !anim || off != 1 {
+		t.Fatalf("first scroll step: off=%d anim=%v", off, anim)
+	}
+
+	// Mid-scroll: offset == 3.
+	off, _, _ = m.step("row", nameLen, budget, t0.Add(marqueeHoldStart+3*marqueeStep))
+	if off != 3 {
+		t.Fatalf("mid scroll: off=%d", off)
+	}
+
+	// Tail reached: offset == maxOffset == 5.
+	off, _, _ = m.step("row", nameLen, budget, t0.Add(marqueeHoldStart+5*marqueeStep+time.Millisecond))
+	if off != 5 {
+		t.Fatalf("tail: off=%d", off)
+	}
+
+	// Hold-end window still pegged at maxOffset.
+	off, _, _ = m.step("row", nameLen, budget, t0.Add(marqueeHoldStart+5*marqueeStep+marqueeHoldEnd/2))
+	if off != 5 {
+		t.Fatalf("hold-end mid: off=%d", off)
+	}
+
+	// After hold-end: snap back to offset 0.
+	off, _, _ = m.step("row", nameLen, budget, t0.Add(marqueeHoldStart+5*marqueeStep+marqueeHoldEnd+time.Millisecond))
+	if off != 0 {
+		t.Fatalf("snap back: off=%d", off)
+	}
+}
+
+func TestMarqueeIDChangeResets(t *testing.T) {
+	var m marqueeState
+	t0 := time.Unix(0, 0)
+	_, _, _ = m.step("a", 10, 5, t0)
+	// Advance well into scroll for row "a".
+	_, _, _ = m.step("a", 10, 5, t0.Add(marqueeHoldStart+3*marqueeStep))
+	// Now focus moves to "b": offset must reset to 0 and phase to hold-start.
+	off, anim, wake := m.step("b", 10, 5, t0.Add(marqueeHoldStart+3*marqueeStep))
+	if off != 0 || !anim || wake != marqueeHoldStart {
+		t.Fatalf("id reset: off=%d anim=%v wake=%v", off, anim, wake)
+	}
+}
+
+func TestMarqueeActive(t *testing.T) {
+	var m marqueeState
+	if m.active() {
+		t.Fatalf("fresh marquee should not be active")
+	}
+	_, _, _ = m.step("row", 10, 5, time.Unix(0, 0))
+	if !m.active() {
+		t.Fatalf("expected active after overflow step")
+	}
+	_, _, _ = m.step("row", 4, 5, time.Unix(0, 0))
+	if m.active() {
+		t.Fatalf("should not be active when name fits")
+	}
+}
+
+func TestMarqueeReset(t *testing.T) {
+	var m marqueeState
+	_, _, _ = m.step("row", 10, 5, time.Unix(0, 0))
+	m.reset()
+	if m.active() {
+		t.Fatalf("expected inactive after reset")
+	}
+	// After reset, stepping the same id starts fresh.
+	off, _, wake := m.step("row", 10, 5, time.Unix(5, 0))
+	if off != 0 || wake != marqueeHoldStart {
+		t.Fatalf("post-reset start: off=%d wake=%v", off, wake)
+	}
+}
+
+func TestFitName(t *testing.T) {
+	cases := []struct {
+		name, in string
+		budget   int
+		want     string
+	}{
+		{"fits", "abc", 5, "abc"},
+		{"exact", "abcde", 5, "abcde"},
+		{"truncate", "abcdef", 5, "abcd…"},
+		{"budget1", "abcdef", 1, "…"},
+		{"budget0", "abc", 0, ""},
+		{"unicode", "αβγδεζη", 4, "αβγ…"},
+	}
+	for _, c := range cases {
+		t.Run(c.name, func(t *testing.T) {
+			got := fitName(c.in, c.budget)
+			if got != c.want {
+				t.Fatalf("fitName(%q, %d) = %q want %q", c.in, c.budget, got, c.want)
+			}
+		})
+	}
+}
+
+func TestMarqueeWindow(t *testing.T) {
+	got := marqueeWindow("abcdefgh", 4, 2)
+	if got != "cdef" {
+		t.Fatalf("window = %q", got)
+	}
+	// Clamp end-of-string overflow.
+	got = marqueeWindow("abcdef", 4, 10)
+	if got != "cdef" {
+		t.Fatalf("clamped window = %q", got)
+	}
+}
+
+func TestClampVisible(t *testing.T) {
+	// Plain string longer than width.
+	if got := clampVisible("abcdef", 3); visibleLen(got) != 3 {
+		t.Fatalf("plain clamp visible = %d (%q)", visibleLen(got), got)
+	}
+	// Already-fitting string is unchanged.
+	if got := clampVisible("abc", 5); got != "abc" {
+		t.Fatalf("unchanged = %q", got)
+	}
+	// SGR-wrapped string: visible portion must be <= width.
+	in := "\x1b[1mhello\x1b[0m world"
+	got := clampVisible(in, 5)
+	if visibleLen(got) != 5 {
+		t.Fatalf("sgr clamp visible = %d (%q)", visibleLen(got), got)
+	}
+}

internal/app/metrics.go

@@ -0,0 +1,462 @@
+package app
+
+import (
+	"context"
+	"encoding/json"
+	"fmt"
+	"os"
+	"path/filepath"
+	"sync/atomic"
+	"time"
+)
+
+// metricsTracker collects per-hot-path counters and timings. All
+// fields are atomic so callers can record from the per-PTY-chunk path
+// without taking a lock. Enabled only when --profile is set.
+//
+// Sampled rates ("X per second", "p99 latency") are not tracked here
+// directly — the snapshotter goroutine writes a row to metrics.jsonl
+// every second, and analysis tools compute rates from the deltas.
+// Aggregate totals are written to metrics.json on shutdown.
+type metricsTracker struct {
+	startedAt time.Time
+
+	// PTY chunk arrival → stdout write pipeline (per OnPTYOut call).
+	ptyChunks      atomic.Int64
+	ptyBytes       atomic.Int64
+	onPTYOutNs     atomic.Int64
+	onPTYOutMaxNs  atomic.Int64
+	onPTYOutDrops  atomic.Int64 // chunks for non-focused children — fast-path returns
+	stdoutWrites   atomic.Int64
+	stdoutBytes    atomic.Int64
+	stdoutNs       atomic.Int64
+	stdoutMaxNs    atomic.Int64
+
+	// Viewport renderer (state-machine over child PTY bytes).
+	renderCalls atomic.Int64
+	renderNs    atomic.Int64
+	renderMaxNs atomic.Int64
+
+	// CGO into libghostty-vt (counted from pumpChild).
+	emuWriteCalls atomic.Int64
+	emuWriteNs    atomic.Int64
+	emuWriteMaxNs atomic.Int64
+	emuTitleCalls atomic.Int64
+	emuTitleNs    atomic.Int64
+	emuTitleSkips atomic.Int64 // OSC-gate fast path — title poll skipped
+
+	// Chrome paint pipeline.
+	sidebarDraws     atomic.Int64
+	sidebarCacheHits atomic.Int64
+	sidebarNs        atomic.Int64
+	sidebarMaxNs     atomic.Int64
+
+	tabbarDraws     atomic.Int64
+	tabbarCacheHits atomic.Int64
+	tabbarNs        atomic.Int64
+
+	statusDraws     atomic.Int64
+	statusCacheHits atomic.Int64
+	statusNs        atomic.Int64
+
+	// Snapshot replay (focus / spawn / nudge).
+	snapshotReplays atomic.Int64
+	snapshotNs      atomic.Int64
+	snapshotMaxNs   atomic.Int64
+
+	// Chrome ticker — distinguishes useful work from idle wakeups.
+	tickerFires      atomic.Int64
+	tickerIdleFires  atomic.Int64 // nothing dirty when the ticker fired
+
+	// Output destination (set when enabled).
+	rowFile *os.File // metrics.jsonl
+	dir     string
+}
+
+// newMetricsTracker creates an enabled tracker writing to <dir>/.
+// Returns nil + nil err if dir is empty (feature off). Caller must
+// call tracker.run(ctx) in a goroutine and tracker.close() at exit.
+func newMetricsTracker(dir string) (*metricsTracker, error) {
+	if dir == "" {
+		return nil, nil
+	}
+	if err := os.MkdirAll(dir, 0o700); err != nil {
+		return nil, err
+	}
+	row, err := os.Create(filepath.Join(dir, "metrics.jsonl"))
+	if err != nil {
+		return nil, err
+	}
+	return &metricsTracker{
+		startedAt: time.Now(),
+		rowFile:   row,
+		dir:       dir,
+	}, nil
+}
+
+// observeMax updates dst to max(dst, v) using a CAS loop. Atomic max
+// isn't a hardware primitive on most CPUs; this is the standard idiom.
+// Spurious wakeups can race but the result settles at the true max.
+func observeMax(dst *atomic.Int64, v int64) {
+	for {
+		old := dst.Load()
+		if v <= old {
+			return
+		}
+		if dst.CompareAndSwap(old, v) {
+			return
+		}
+	}
+}
+
+// recordPTYOut is called once at the end of each OnPTYOut invocation.
+// `dur` is the full per-chunk wall time (renderer + stdout + chrome
+// signals); `bytes` is the chunk's byte count.
+func (m *metricsTracker) recordPTYOut(dur time.Duration, bytes int) {
+	if m == nil {
+		return
+	}
+	m.ptyChunks.Add(1)
+	m.ptyBytes.Add(int64(bytes))
+	ns := dur.Nanoseconds()
+	m.onPTYOutNs.Add(ns)
+	observeMax(&m.onPTYOutMaxNs, ns)
+}
+
+func (m *metricsTracker) recordPTYOutDrop() {
+	if m == nil {
+		return
+	}
+	m.onPTYOutDrops.Add(1)
+}
+
+func (m *metricsTracker) recordRender(dur time.Duration) {
+	if m == nil {
+		return
+	}
+	m.renderCalls.Add(1)
+	ns := dur.Nanoseconds()
+	m.renderNs.Add(ns)
+	observeMax(&m.renderMaxNs, ns)
+}
+
+func (m *metricsTracker) recordStdout(dur time.Duration, bytes int) {
+	if m == nil {
+		return
+	}
+	m.stdoutWrites.Add(1)
+	m.stdoutBytes.Add(int64(bytes))
+	ns := dur.Nanoseconds()
+	m.stdoutNs.Add(ns)
+	observeMax(&m.stdoutMaxNs, ns)
+}
+
+func (m *metricsTracker) recordEmuWrite(dur time.Duration) {
+	if m == nil {
+		return
+	}
+	m.emuWriteCalls.Add(1)
+	ns := dur.Nanoseconds()
+	m.emuWriteNs.Add(ns)
+	observeMax(&m.emuWriteMaxNs, ns)
+}
+
+func (m *metricsTracker) recordEmuTitle(dur time.Duration, skipped bool) {
+	if m == nil {
+		return
+	}
+	if skipped {
+		m.emuTitleSkips.Add(1)
+		return
+	}
+	m.emuTitleCalls.Add(1)
+	m.emuTitleNs.Add(dur.Nanoseconds())
+}
+
+func (m *metricsTracker) recordSidebar(dur time.Duration, cacheHit bool) {
+	if m == nil {
+		return
+	}
+	m.sidebarDraws.Add(1)
+	if cacheHit {
+		m.sidebarCacheHits.Add(1)
+	}
+	ns := dur.Nanoseconds()
+	m.sidebarNs.Add(ns)
+	observeMax(&m.sidebarMaxNs, ns)
+}
+
+func (m *metricsTracker) recordTabbar(dur time.Duration, cacheHit bool) {
+	if m == nil {
+		return
+	}
+	m.tabbarDraws.Add(1)
+	if cacheHit {
+		m.tabbarCacheHits.Add(1)
+	}
+	m.tabbarNs.Add(dur.Nanoseconds())
+}
+
+func (m *metricsTracker) recordStatus(dur time.Duration, cacheHit bool) {
+	if m == nil {
+		return
+	}
+	m.statusDraws.Add(1)
+	if cacheHit {
+		m.statusCacheHits.Add(1)
+	}
+	m.statusNs.Add(dur.Nanoseconds())
+}
+
+func (m *metricsTracker) recordSnapshot(dur time.Duration) {
+	if m == nil {
+		return
+	}
+	m.snapshotReplays.Add(1)
+	ns := dur.Nanoseconds()
+	m.snapshotNs.Add(ns)
+	observeMax(&m.snapshotMaxNs, ns)
+}
+
+func (m *metricsTracker) recordTickerFire(didWork bool) {
+	if m == nil {
+		return
+	}
+	m.tickerFires.Add(1)
+	if !didWork {
+		m.tickerIdleFires.Add(1)
+	}
+}
+
+// snapshot captures the tracker's current state as a JSON-serialisable
+// map. Suitable for both the per-second JSONL row and the final
+// metrics.json aggregate.
+type metricsSnapshot struct {
+	WallSeconds   float64 `json:"wall_seconds"`
+	PTYChunks     int64   `json:"pty_chunks"`
+	PTYBytes      int64   `json:"pty_bytes"`
+	OnPTYOutNs    int64   `json:"on_pty_out_ns_total"`
+	OnPTYOutMaxNs int64   `json:"on_pty_out_ns_max"`
+	OnPTYOutDrops int64   `json:"on_pty_out_drops"`
+	StdoutWrites  int64   `json:"stdout_writes"`
+	StdoutBytes   int64   `json:"stdout_bytes"`
+	StdoutNs      int64   `json:"stdout_ns_total"`
+	StdoutMaxNs   int64   `json:"stdout_ns_max"`
+
+	RenderCalls int64 `json:"render_calls"`
+	RenderNs    int64 `json:"render_ns_total"`
+	RenderMaxNs int64 `json:"render_ns_max"`
+
+	EmuWriteCalls int64 `json:"emu_write_calls"`
+	EmuWriteNs    int64 `json:"emu_write_ns_total"`
+	EmuWriteMaxNs int64 `json:"emu_write_ns_max"`
+	EmuTitleCalls int64 `json:"emu_title_calls"`
+	EmuTitleNs    int64 `json:"emu_title_ns_total"`
+	EmuTitleSkips int64 `json:"emu_title_skips"`
+
+	SidebarDraws     int64 `json:"sidebar_draws"`
+	SidebarCacheHits int64 `json:"sidebar_cache_hits"`
+	SidebarNs        int64 `json:"sidebar_ns_total"`
+	SidebarMaxNs     int64 `json:"sidebar_ns_max"`
+
+	TabbarDraws     int64 `json:"tabbar_draws"`
+	TabbarCacheHits int64 `json:"tabbar_cache_hits"`
+	TabbarNs        int64 `json:"tabbar_ns_total"`
+
+	StatusDraws     int64 `json:"status_draws"`
+	StatusCacheHits int64 `json:"status_cache_hits"`
+	StatusNs        int64 `json:"status_ns_total"`
+
+	SnapshotReplays int64 `json:"snapshot_replays"`
+	SnapshotNs      int64 `json:"snapshot_ns_total"`
+	SnapshotMaxNs   int64 `json:"snapshot_ns_max"`
+
+	TickerFires     int64 `json:"ticker_fires"`
+	TickerIdleFires int64 `json:"ticker_idle_fires"`
+
+	// Derived rates (computed at snapshot time so consumers don't have
+	// to). All "per_second" values are averaged over wall_seconds.
+	PTYChunksPerSec      float64 `json:"pty_chunks_per_sec"`
+	PTYBytesPerSec       float64 `json:"pty_bytes_per_sec"`
+	OnPTYOutMeanUs       float64 `json:"on_pty_out_mean_us"`
+	StdoutMeanUs         float64 `json:"stdout_mean_us"`
+	EmuWriteMeanUs       float64 `json:"emu_write_mean_us"`
+	SidebarMeanUs        float64 `json:"sidebar_mean_us"`
+	SidebarCacheHitRate  float64 `json:"sidebar_cache_hit_rate"`
+	TabbarCacheHitRate   float64 `json:"tabbar_cache_hit_rate"`
+	StatusCacheHitRate   float64 `json:"status_cache_hit_rate"`
+	EmuTitleSkipRate     float64 `json:"emu_title_skip_rate"`
+	TickerIdleRate       float64 `json:"ticker_idle_rate"`
+	Timestamp            string  `json:"timestamp"`
+}
+
+func (m *metricsTracker) snapshotNow() metricsSnapshot {
+	wall := time.Since(m.startedAt).Seconds()
+	if wall <= 0 {
+		wall = 1
+	}
+	chunks := m.ptyChunks.Load()
+	bytes := m.ptyBytes.Load()
+	onptyTotal := m.onPTYOutNs.Load()
+	stdW := m.stdoutWrites.Load()
+	stdNs := m.stdoutNs.Load()
+	emuW := m.emuWriteCalls.Load()
+	emuWNs := m.emuWriteNs.Load()
+	sbDraws := m.sidebarDraws.Load()
+	sbHits := m.sidebarCacheHits.Load()
+	sbNs := m.sidebarNs.Load()
+	tbDraws := m.tabbarDraws.Load()
+	tbHits := m.tabbarCacheHits.Load()
+	stDraws := m.statusDraws.Load()
+	stHits := m.statusCacheHits.Load()
+	emuTC := m.emuTitleCalls.Load()
+	emuTS := m.emuTitleSkips.Load()
+	tickerF := m.tickerFires.Load()
+	tickerI := m.tickerIdleFires.Load()
+
+	div := func(num, denom int64) float64 {
+		if denom == 0 {
+			return 0
+		}
+		return float64(num) / float64(denom)
+	}
+
+	return metricsSnapshot{
+		WallSeconds:   wall,
+		PTYChunks:     chunks,
+		PTYBytes:      bytes,
+		OnPTYOutNs:    onptyTotal,
+		OnPTYOutMaxNs: m.onPTYOutMaxNs.Load(),
+		OnPTYOutDrops: m.onPTYOutDrops.Load(),
+		StdoutWrites:  stdW,
+		StdoutBytes:   m.stdoutBytes.Load(),
+		StdoutNs:      stdNs,
+		StdoutMaxNs:   m.stdoutMaxNs.Load(),
+
+		RenderCalls: m.renderCalls.Load(),
+		RenderNs:    m.renderNs.Load(),
+		RenderMaxNs: m.renderMaxNs.Load(),
+
+		EmuWriteCalls: emuW,
+		EmuWriteNs:    emuWNs,
+		EmuWriteMaxNs: m.emuWriteMaxNs.Load(),
+		EmuTitleCalls: emuTC,
+		EmuTitleNs:    m.emuTitleNs.Load(),
+		EmuTitleSkips: emuTS,
+
+		SidebarDraws:     sbDraws,
+		SidebarCacheHits: sbHits,
+		SidebarNs:        sbNs,
+		SidebarMaxNs:     m.sidebarMaxNs.Load(),
+
+		TabbarDraws:     tbDraws,
+		TabbarCacheHits: tbHits,
+		TabbarNs:        m.tabbarNs.Load(),
+
+		StatusDraws:     stDraws,
+		StatusCacheHits: stHits,
+		StatusNs:        m.statusNs.Load(),
+
+		SnapshotReplays: m.snapshotReplays.Load(),
+		SnapshotNs:      m.snapshotNs.Load(),
+		SnapshotMaxNs:   m.snapshotMaxNs.Load(),
+
+		TickerFires:     tickerF,
+		TickerIdleFires: tickerI,
+
+		PTYChunksPerSec:     float64(chunks) / wall,
+		PTYBytesPerSec:      float64(bytes) / wall,
+		OnPTYOutMeanUs:      div(onptyTotal/1000, chunks),
+		StdoutMeanUs:        div(stdNs/1000, stdW),
+		EmuWriteMeanUs:      div(emuWNs/1000, emuW),
+		SidebarMeanUs:       div(sbNs/1000, sbDraws),
+		SidebarCacheHitRate: div(sbHits, sbDraws),
+		TabbarCacheHitRate:  div(tbHits, tbDraws),
+		StatusCacheHitRate:  div(stHits, stDraws),
+		EmuTitleSkipRate:    div(emuTS, emuTC+emuTS),
+		TickerIdleRate:      div(tickerI, tickerF),
+		Timestamp:           time.Now().Format(time.RFC3339Nano),
+	}
+}
+
+// run is the snapshotter goroutine: write a JSONL row every second
+// until ctx is cancelled. Stops cleanly without flushing partial
+// rows.
+func (m *metricsTracker) run(ctx context.Context) {
+	if m == nil {
+		return
+	}
+	enc := json.NewEncoder(m.rowFile)
+	ticker := time.NewTicker(time.Second)
+	defer ticker.Stop()
+	for {
+		select {
+		case <-ctx.Done():
+			return
+		case <-ticker.C:
+			snap := m.snapshotNow()
+			_ = enc.Encode(snap)
+		}
+	}
+}
+
+// close writes the final aggregate snapshot to metrics.json + a
+// short human-readable summary.txt, then closes the row file. Safe
+// to call on a nil receiver.
+func (m *metricsTracker) close() {
+	if m == nil {
+		return
+	}
+	snap := m.snapshotNow()
+	if f, err := os.Create(filepath.Join(m.dir, "metrics.json")); err == nil {
+		enc := json.NewEncoder(f)
+		enc.SetIndent("", "  ")
+		_ = enc.Encode(snap)
+		_ = f.Close()
+	}
+	if f, err := os.Create(filepath.Join(m.dir, "summary.txt")); err == nil {
+		writeSummary(f, snap)
+		_ = f.Close()
+	}
+	if m.rowFile != nil {
+		_ = m.rowFile.Close()
+		m.rowFile = nil
+	}
+}
+
+// writeSummary renders a brief human-readable digest of a snapshot.
+// Designed for `cat summary.txt` after a session — quick orientation
+// before diving into metrics.json / pprof.
+func writeSummary(w *os.File, s metricsSnapshot) {
+	fmt.Fprintf(w, "patterm performance summary\n")
+	fmt.Fprintf(w, "===========================\n\n")
+	fmt.Fprintf(w, "session length:        %.1fs\n", s.WallSeconds)
+	fmt.Fprintf(w, "pty chunks:            %d  (%.1f /s)\n", s.PTYChunks, s.PTYChunksPerSec)
+	fmt.Fprintf(w, "pty bytes:             %d  (%.0f /s, %.1f KiB/s)\n",
+		s.PTYBytes, s.PTYBytesPerSec, s.PTYBytesPerSec/1024)
+	fmt.Fprintf(w, "pty chunks dropped:    %d  (focus not on caller — fast-path return)\n", s.OnPTYOutDrops)
+	fmt.Fprintf(w, "\n")
+	fmt.Fprintf(w, "OnPTYOut mean:         %.1fµs   max: %.1fms\n",
+		s.OnPTYOutMeanUs, float64(s.OnPTYOutMaxNs)/1e6)
+	fmt.Fprintf(w, "viewport.Render calls: %d  total %.1fms  max %.1fms\n",
+		s.RenderCalls, float64(s.RenderNs)/1e6, float64(s.RenderMaxNs)/1e6)
+	fmt.Fprintf(w, "stdout writes:         %d  mean %.1fµs  max %.1fms  bytes %d\n",
+		s.StdoutWrites, s.StdoutMeanUs, float64(s.StdoutMaxNs)/1e6, s.StdoutBytes)
+	fmt.Fprintf(w, "\n")
+	fmt.Fprintf(w, "emulator.Write (cgo):  %d  mean %.1fµs  max %.1fms\n",
+		s.EmuWriteCalls, s.EmuWriteMeanUs, float64(s.EmuWriteMaxNs)/1e6)
+	fmt.Fprintf(w, "emulator.Title polls:  %d real, %d gated   skip rate %.1f%%\n",
+		s.EmuTitleCalls, s.EmuTitleSkips, s.EmuTitleSkipRate*100)
+	fmt.Fprintf(w, "\n")
+	fmt.Fprintf(w, "sidebar draws:         %d  mean %.1fµs  max %.1fms  cache-hit %.1f%%\n",
+		s.SidebarDraws, s.SidebarMeanUs, float64(s.SidebarMaxNs)/1e6, s.SidebarCacheHitRate*100)
+	fmt.Fprintf(w, "tabbar draws:          %d  cache-hit %.1f%%\n",
+		s.TabbarDraws, s.TabbarCacheHitRate*100)
+	fmt.Fprintf(w, "status draws:          %d  cache-hit %.1f%%\n",
+		s.StatusDraws, s.StatusCacheHitRate*100)
+	fmt.Fprintf(w, "snapshot replays:      %d  total %.1fms  max %.1fms\n",
+		s.SnapshotReplays, float64(s.SnapshotNs)/1e6, float64(s.SnapshotMaxNs)/1e6)
+	fmt.Fprintf(w, "\n")
+	fmt.Fprintf(w, "chrome ticker:         %d fires, %d idle   idle rate %.1f%%\n",
+		s.TickerFires, s.TickerIdleFires, s.TickerIdleRate*100)
+}

internal/app/metrics_test.go

@@ -0,0 +1,116 @@
+package app
+
+import (
+	"encoding/json"
+	"os"
+	"path/filepath"
+	"testing"
+	"time"
+)
+
+func TestMetricsTrackerDisabledByEmptyDir(t *testing.T) {
+	m, err := newMetricsTracker("")
+	if err != nil {
+		t.Fatalf("newMetricsTracker(\"\") err: %v", err)
+	}
+	if m != nil {
+		t.Fatalf("expected nil tracker for empty dir, got %v", m)
+	}
+}
+
+func TestMetricsTrackerRecordsAndWrites(t *testing.T) {
+	dir := t.TempDir()
+	m, err := newMetricsTracker(dir)
+	if err != nil {
+		t.Fatalf("newMetricsTracker: %v", err)
+	}
+	if m == nil {
+		t.Fatal("expected non-nil tracker")
+	}
+
+	m.recordPTYOut(2*time.Millisecond, 1024)
+	m.recordPTYOut(5*time.Millisecond, 4096)
+	m.recordRender(800 * time.Microsecond)
+	m.recordStdout(300*time.Microsecond, 1100)
+	m.recordEmuWrite(150 * time.Microsecond)
+	m.recordEmuTitle(0, true)
+	m.recordEmuTitle(20*time.Microsecond, false)
+	m.recordSidebar(100*time.Microsecond, true)
+	m.recordSidebar(900*time.Microsecond, false)
+	m.recordTabbar(50*time.Microsecond, true)
+	m.recordStatus(40*time.Microsecond, true)
+	m.recordSnapshot(2 * time.Millisecond)
+	m.recordTickerFire(false)
+	m.recordTickerFire(true)
+	m.recordPTYOutDrop()
+
+	m.close()
+
+	// metrics.json should exist and parse, and reflect what we recorded.
+	raw, err := os.ReadFile(filepath.Join(dir, "metrics.json"))
+	if err != nil {
+		t.Fatalf("read metrics.json: %v", err)
+	}
+	var snap metricsSnapshot
+	if err := json.Unmarshal(raw, &snap); err != nil {
+		t.Fatalf("parse metrics.json: %v", err)
+	}
+	if snap.PTYChunks != 2 {
+		t.Errorf("PTYChunks = %d, want 2", snap.PTYChunks)
+	}
+	if snap.PTYBytes != 5120 {
+		t.Errorf("PTYBytes = %d, want 5120", snap.PTYBytes)
+	}
+	if snap.OnPTYOutMaxNs != (5 * time.Millisecond).Nanoseconds() {
+		t.Errorf("OnPTYOutMaxNs = %d, want %d",
+			snap.OnPTYOutMaxNs, (5 * time.Millisecond).Nanoseconds())
+	}
+	if snap.SidebarDraws != 2 {
+		t.Errorf("SidebarDraws = %d, want 2", snap.SidebarDraws)
+	}
+	if snap.SidebarCacheHits != 1 {
+		t.Errorf("SidebarCacheHits = %d, want 1", snap.SidebarCacheHits)
+	}
+	if snap.SidebarCacheHitRate != 0.5 {
+		t.Errorf("SidebarCacheHitRate = %v, want 0.5", snap.SidebarCacheHitRate)
+	}
+	if snap.EmuTitleCalls != 1 || snap.EmuTitleSkips != 1 {
+		t.Errorf("emu title accounting: calls=%d skips=%d, want 1/1",
+			snap.EmuTitleCalls, snap.EmuTitleSkips)
+	}
+	if snap.TickerFires != 2 || snap.TickerIdleFires != 1 {
+		t.Errorf("ticker accounting: fires=%d idle=%d, want 2/1",
+			snap.TickerFires, snap.TickerIdleFires)
+	}
+	if snap.OnPTYOutDrops != 1 {
+		t.Errorf("OnPTYOutDrops = %d, want 1", snap.OnPTYOutDrops)
+	}
+
+	// summary.txt should also be present and non-empty.
+	info, err := os.Stat(filepath.Join(dir, "summary.txt"))
+	if err != nil {
+		t.Fatalf("stat summary.txt: %v", err)
+	}
+	if info.Size() == 0 {
+		t.Fatal("summary.txt is empty")
+	}
+}
+
+func TestMetricsTrackerNilSafe(t *testing.T) {
+	// Every record* method must be safe to call on a nil receiver
+	// because the hot paths use that to avoid an enabled-check.
+	var m *metricsTracker
+	m.recordPTYOut(time.Millisecond, 100)
+	m.recordPTYOutDrop()
+	m.recordRender(time.Microsecond)
+	m.recordStdout(time.Microsecond, 50)
+	m.recordEmuWrite(time.Microsecond)
+	m.recordEmuTitle(time.Microsecond, false)
+	m.recordEmuTitle(0, true)
+	m.recordSidebar(time.Microsecond, true)
+	m.recordTabbar(time.Microsecond, false)
+	m.recordStatus(time.Microsecond, true)
+	m.recordSnapshot(time.Microsecond)
+	m.recordTickerFire(true)
+	m.close()
+}

internal/app/palette_context_test.go

@@ -0,0 +1,233 @@
+package app
+
+import (
+	"os/exec"
+	"testing"
+	"time"
+
+	"github.com/hjbdev/patterm/internal/preset"
+)
+
+// makeFakeChild builds a Child with just enough state for the palette
+// to render it. We don't start a PTY — the palette only reads ID,
+// Name, Kind, and Status() which all work without one.
+func makeFakeChild(id, name string, kind ChildKind) *Child {
+	c := &Child{ID: id, Name: name, Kind: kind}
+	st := StatusRunning
+	c.status.Store(&st)
+	return c
+}
+
+// findAction scans p.items and returns the first paletteAction.kind
+// matching want, or "" if not found.
+func findItem(p *paletteState, want string) (int, *paletteItem) {
+	for i := range p.items {
+		if p.items[i].action.kind == want {
+			return i, &p.items[i]
+		}
+	}
+	return -1, nil
+}
+
+func TestContextItemsScratchpad(t *testing.T) {
+	p := newPalette(nil, "", "notes.md", preset.Set{})
+	// With the dashed section header gone, pad-edit is the first row;
+	// pad-rename-form follows, with destructive pad-delete last in the
+	// Focused section.
+	if i, _ := findItem(p, "pad-edit"); i != 0 {
+		t.Fatalf("pad-edit at %d; want 0", i)
+	}
+	if _, it := findItem(p, "pad-rename-form"); it == nil || it.action.padName != "notes.md" {
+		t.Fatalf("pad-rename-form missing or wrong padName: %+v", it)
+	}
+	if i, _ := findItem(p, "pad-delete"); i < 0 {
+		t.Fatalf("pad-delete missing")
+	}
+	// No focused child → no agent/proc context items.
+	if i, _ := findItem(p, "agent-rename-form"); i != -1 {
+		t.Fatalf("agent items leaked: index %d", i)
+	}
+	if i, _ := findItem(p, "proc-rename-form"); i != -1 {
+		t.Fatalf("proc items leaked: index %d", i)
+	}
+}
+
+func TestContextItemsAgent(t *testing.T) {
+	c := makeFakeChild("aid", "codex", KindAgent)
+	p := newPalette([]*Child{c}, "aid", "", preset.Set{})
+	if _, it := findItem(p, "agent-rename-form"); it == nil || it.action.childID != "aid" {
+		t.Fatalf("agent-rename-form missing or wrong: %+v", it)
+	}
+	if _, it := findItem(p, "agent-close"); it == nil || it.action.childID != "aid" {
+		t.Fatalf("agent-close missing or wrong: %+v", it)
+	}
+	// agent context never surfaces proc items.
+	if i, _ := findItem(p, "proc-rename-form"); i != -1 {
+		t.Fatalf("proc items leaked into agent context: index %d", i)
+	}
+	if i, _ := findItem(p, "pad-delete"); i != -1 {
+		t.Fatalf("pad items leaked into agent context")
+	}
+}
+
+func TestContextItemsProcess(t *testing.T) {
+	c := makeFakeChild("pid", "devserver", KindCommand)
+	p := newPalette([]*Child{c}, "pid", "", preset.Set{})
+	for _, kind := range []string{"proc-rename-form", "proc-delete", "proc-stop", "proc-restart"} {
+		if _, it := findItem(p, kind); it == nil {
+			t.Fatalf("missing proc context item: %s", kind)
+		}
+	}
+	if i, _ := findItem(p, "agent-rename-form"); i != -1 {
+		t.Fatalf("agent items leaked into process context")
+	}
+}
+
+func TestContextItemsTerminalUsesCloseNotStop(t *testing.T) {
+	c := makeFakeChild("tid", "terminal", KindTerminal)
+	p := newPalette([]*Child{c}, "tid", "", preset.Set{})
+	if _, it := findItem(p, "proc-stop"); it == nil || it.label != "Close" {
+		t.Fatalf("terminal close row missing or mislabelled: %+v", it)
+	}
+	if _, it := findItem(p, "proc-restart"); it == nil {
+		t.Fatalf("terminal restart row missing")
+	}
+	if i, _ := findItem(p, "proc-delete"); i != -1 {
+		t.Fatalf("terminal should not show a separate delete/close row, found at %d", i)
+	}
+	for i, it := range p.items {
+		if it.label == "Stop" {
+			t.Fatalf("terminal should not show Stop row, found at %d", i)
+		}
+	}
+}
+
+func TestContextItemsAppearAboveSwitch(t *testing.T) {
+	// Two children so there's still a non-focused switch entry to compare
+	// against (the focused child is suppressed from the Open section).
+	focused := makeFakeChild("pid", "devserver", KindCommand)
+	other := makeFakeChild("oid", "worker", KindCommand)
+	p := newPalette([]*Child{focused, other}, "pid", "", preset.Set{})
+	procIdx, _ := findItem(p, "proc-rename-form")
+	switchIdx, _ := findItem(p, "switch")
+	if procIdx < 0 || switchIdx < 0 {
+		t.Fatalf("missing items: proc=%d switch=%d", procIdx, switchIdx)
+	}
+	if procIdx > switchIdx {
+		t.Fatalf("proc context item at %d came after switch at %d", procIdx, switchIdx)
+	}
+}
+
+func TestContextItemsNoFocusNoExtras(t *testing.T) {
+	p := newPalette(nil, "", "", preset.Set{})
+	for _, kind := range []string{
+		"pad-delete", "pad-rename-form", "pad-edit",
+		"agent-rename-form", "agent-close",
+		"proc-rename-form", "proc-delete", "proc-stop", "proc-restart",
+	} {
+		if i, _ := findItem(p, kind); i != -1 {
+			t.Fatalf("unexpected context item %s with no focus (idx=%d)", kind, i)
+		}
+	}
+}
+
+// Renaming a scratchpad via Enter should open the rename form, accept
+// typed input, and emit a pad-rename-submit with the new name.
+func TestRenamePadFormCommits(t *testing.T) {
+	p := newPalette(nil, "", "notes.md", preset.Set{})
+	idx, _ := findItem(p, "pad-rename-form")
+	if idx < 0 {
+		t.Fatalf("pad-rename-form missing")
+	}
+	p.cursor = idx
+	// Open the form.
+	_, done, _ := p.handleInput([]byte("\r"), 0)
+	if done {
+		t.Fatalf("opening rename form closed palette")
+	}
+	if p.mode != paletteModeRenameForm || p.renameForm == nil {
+		t.Fatalf("mode=%v form=%v after open", p.mode, p.renameForm)
+	}
+	if string(p.renameForm.name) != "notes.md" {
+		t.Fatalf("prefill = %q", string(p.renameForm.name))
+	}
+	// Clear and type a new name.
+	_, _, _ = p.handleInput([]byte{0x15}, 0) // Ctrl-U
+	if len(p.renameForm.name) != 0 {
+		t.Fatalf("Ctrl-U didn't clear: %q", string(p.renameForm.name))
+	}
+	for _, b := range []byte("brief.md") {
+		_, _, _ = p.handleInput([]byte{b}, 0)
+	}
+	action, done, _ := p.handleInput([]byte("\r"), 0)
+	if !done || action.kind != "pad-rename-submit" {
+		t.Fatalf("submit didn't fire: action=%+v done=%v", action, done)
+	}
+	if action.padName != "notes.md" || action.newName != "brief.md" {
+		t.Fatalf("submit payload = %+v", action)
+	}
+}
+
+func TestRenameProcessFormPrefillsCurrentName(t *testing.T) {
+	c := makeFakeChild("pid", "devserver", KindCommand)
+	p := newPalette([]*Child{c}, "pid", "", preset.Set{})
+	idx, _ := findItem(p, "proc-rename-form")
+	if idx < 0 {
+		t.Fatalf("proc-rename-form missing")
+	}
+	p.cursor = idx
+	_, _, _ = p.handleInput([]byte("\r"), 0)
+	if p.renameForm == nil || string(p.renameForm.name) != "devserver" {
+		t.Fatalf("prefill = %v", p.renameForm)
+	}
+	if p.renameForm.subject != "proc" || p.renameForm.target != "pid" {
+		t.Fatalf("form target/subject wrong: %+v", p.renameForm)
+	}
+}
+
+func TestRenameFormEscCancels(t *testing.T) {
+	p := newPalette(nil, "", "notes.md", preset.Set{})
+	p.mode = paletteModeRenameForm
+	p.renameForm = &renameForm{name: []rune("x"), subject: "pad", target: "notes.md"}
+	action, done, _ := p.handleInput([]byte{0x1b}, 0)
+	if !done || action.kind != "cancel" {
+		t.Fatalf("ESC didn't cancel: action=%+v done=%v", action, done)
+	}
+}
+
+func TestRenameFormEmptySubmitCancels(t *testing.T) {
+	p := newPalette(nil, "", "notes.md", preset.Set{})
+	p.mode = paletteModeRenameForm
+	p.renameForm = &renameForm{name: []rune("   "), subject: "pad", target: "notes.md"}
+	action, done, _ := p.handleInput([]byte("\r"), 0)
+	if !done || action.kind != "cancel" {
+		t.Fatalf("empty submit didn't cancel: action=%+v done=%v", action, done)
+	}
+}
+
+// TestPadEditDoesNotBlock guards the "fire-and-forget exec" contract:
+// handlePadEdit must Start() the editor and return promptly, not Wait()
+// on it. We substitute a slow command (`sleep 30`) via PATH and ensure
+// the action returns well under a second.
+func TestPadEditDoesNotBlock(t *testing.T) {
+	if _, err := exec.LookPath("sleep"); err != nil {
+		t.Skip("no sleep on PATH")
+	}
+	// Verify the action runs through exec.Command/Start in well under a
+	// second by directly invoking the same primitive handlePadEdit uses.
+	cmd := exec.Command("sleep", "30")
+	start := time.Now()
+	if err := cmd.Start(); err != nil {
+		t.Fatalf("start: %v", err)
+	}
+	if cmd.Process != nil {
+		_ = cmd.Process.Release()
+		// Best-effort cleanup so the test doesn't leave a sleeping
+		// process behind. Release() detaches from the parent so a
+		// follow-up kill is the only way to reap it deterministically.
+		_ = cmd.Process.Kill()
+	}
+	if elapsed := time.Since(start); elapsed > 500*time.Millisecond {
+		t.Fatalf("exec.Start took %v — handlePadEdit would block the TUI", elapsed)
+	}
+}

internal/app/palette_input_test.go

@@ -7,7 +7,7 @@ import (
 )
 
 func newTestPalette() *paletteState {
-	return newPalette(nil, "", preset.Set{})
+	return newPalette(nil, "", "", preset.Set{})
 }
 
 func TestPaletteIgnoresKittyReleaseEvent(t *testing.T) {
@@ -47,42 +47,56 @@ func TestPaletteBareEscCancels(t *testing.T) {
 	}
 }
 
+// firstSelectable returns the lowest item index whose action is
+// selectable (not a section header), or -1 if the palette has no
+// selectable rows.
+func firstSelectable(p *paletteState) int {
+	for i, it := range p.items {
+		if it.action.kind != "header" {
+			return i
+		}
+	}
+	return -1
+}
+
 func TestPaletteKittyArrowsNavigate(t *testing.T) {
 	pr := []*preset.Preset{{Name: "a"}, {Name: "b"}, {Name: "c"}}
-	p := newPalette(nil, "", preset.Set{Agents: pr})
-	if p.cursor != 0 {
-		t.Fatalf("initial cursor %d", p.cursor)
+	p := newPalette(nil, "", "", preset.Set{Agents: pr})
+	first := firstSelectable(p)
+	if first < 0 || p.cursor != first {
+		t.Fatalf("initial cursor %d, want first selectable %d", p.cursor, first)
 	}
 	// Kitty functional Down arrow.
 	_, _, adv := p.handleInput([]byte("\x1b[57353u"), 0)
 	if adv != 8 {
 		t.Fatalf("advance %d", adv)
 	}
-	if p.cursor != 1 {
-		t.Fatalf("cursor %d after Down, want 1", p.cursor)
+	if p.cursor != first+1 {
+		t.Fatalf("cursor %d after Down, want %d", p.cursor, first+1)
 	}
 	// Kitty functional Up arrow.
 	_, _, _ = p.handleInput([]byte("\x1b[57352u"), 0)
-	if p.cursor != 0 {
-		t.Fatalf("cursor %d after Up, want 0", p.cursor)
+	if p.cursor != first {
+		t.Fatalf("cursor %d after Up, want %d", p.cursor, first)
 	}
 }
 
 func TestPaletteLegacyArrowsStillWork(t *testing.T) {
 	pr := []*preset.Preset{{Name: "a"}, {Name: "b"}}
-	p := newPalette(nil, "", preset.Set{Agents: pr})
+	p := newPalette(nil, "", "", preset.Set{Agents: pr})
+	first := firstSelectable(p)
 	_, _, adv := p.handleInput([]byte("\x1b[B"), 0)
 	if adv != 3 {
 		t.Fatalf("advance %d", adv)
 	}
-	if p.cursor != 1 {
-		t.Fatalf("cursor %d, want 1", p.cursor)
+	if p.cursor != first+1 {
+		t.Fatalf("cursor %d, want %d", p.cursor, first+1)
 	}
 }
 
 func TestPaletteKittyEnterAccepts(t *testing.T) {
 	pr := []*preset.Preset{{Name: "x"}}
-	p := newPalette(nil, "", preset.Set{Agents: pr})
+	p := newPalette(nil, "", "", preset.Set{Agents: pr})
 	action, done, _ := p.handleInput([]byte("\x1b[13u"), 0)
 	if !done || action.kind != "spawn-agent" {
 		t.Fatalf("Enter via CSI u didn't accept: action=%+v done=%v", action, done)
@@ -112,7 +126,7 @@ func TestPaletteLegacyPrintableTypes(t *testing.T) {
 // non-empty command line emits the submit action with relaunch reflecting
 // the checkbox state.
 func TestPaletteSpawnProcessFormFlow(t *testing.T) {
-	p := newPalette(nil, "", preset.Set{})
+	p := newPalette(nil, "", "", preset.Set{})
 	// The "Spawn process…" entry is the only non-Quit item with an
 	// empty preset list. Locate its index by scanning items.
 	idx := -1
@@ -165,7 +179,7 @@ func TestPaletteSpawnProcessFormFlow(t *testing.T) {
 }
 
 func TestPaletteSpawnProcessFormEmptyCommandCancels(t *testing.T) {
-	p := newPalette(nil, "", preset.Set{})
+	p := newPalette(nil, "", "", preset.Set{})
 	p.mode = paletteModeSpawnForm
 	p.form = &spawnProcessForm{}
 	action, done, _ := p.handleInput([]byte("\r"), 0)
@@ -175,7 +189,7 @@ func TestPaletteSpawnProcessFormEmptyCommandCancels(t *testing.T) {
 }
 
 func TestPaletteSpawnProcessFormEscCancels(t *testing.T) {
-	p := newPalette(nil, "", preset.Set{})
+	p := newPalette(nil, "", "", preset.Set{})
 	p.mode = paletteModeSpawnForm
 	p.form = &spawnProcessForm{cmd: []rune("x")}
 	action, done, _ := p.handleInput([]byte{0x1b}, 0)

internal/app/palette_ux_test.go

@@ -0,0 +1,484 @@
+package app
+
+import (
+	"bytes"
+	"strings"
+	"testing"
+
+	"github.com/hjbdev/patterm/internal/preset"
+)
+
+// -- Phase 1: naming & dropped global Close list ---------------------
+
+func TestPaletteVerbsAreUnified(t *testing.T) {
+	procs := []*preset.Preset{{Name: "dev"}}
+	agents := []*preset.Preset{{Name: "claude"}}
+	p := newPalette(nil, "", "", preset.Set{Agents: agents, Processes: procs})
+	gotLabels := make([]string, 0, len(p.items))
+	for _, it := range p.items {
+		if it.action.kind == "header" {
+			continue
+		}
+		gotLabels = append(gotLabels, it.label)
+	}
+	joined := strings.Join(gotLabels, "\n")
+
+	mustContain := []string{
+		"Spawn agent: claude",
+		"Spawn process: dev",
+		"Spawn terminal",
+		"Spawn process… (custom)",
+	}
+	for _, want := range mustContain {
+		if !strings.Contains(joined, want) {
+			t.Errorf("missing unified-verb label %q in:\n%s", want, joined)
+		}
+	}
+	// The pre-overhaul verb forms must not appear anywhere.
+	mustNotContain := []string{"Run process:", "New Terminal", "Spawn process… (custom)"}
+	for _, bad := range mustNotContain {
+		if strings.Contains(joined, bad) {
+			t.Errorf("leftover legacy verb %q present in:\n%s", bad, joined)
+		}
+	}
+}
+
+func TestPaletteDropsGlobalCloseList(t *testing.T) {
+	c1 := makeFakeChild("a", "claude", KindAgent)
+	c2 := makeFakeChild("b", "dev", KindCommand)
+	p := newPalette([]*Child{c1, c2}, "", "", preset.Set{})
+	// No focus → no Focused context, so no "kill" / "agent-close" /
+	// "proc-stop" rows should exist at all.
+	for _, kind := range []string{"kill", "agent-close", "proc-stop", "proc-delete"} {
+		if i, _ := findItem(p, kind); i != -1 {
+			t.Fatalf("kind %q present at %d; global Close list should be gone", kind, i)
+		}
+	}
+}
+
+// -- Phase 2: section headers and cursor skip ------------------------
+
+func TestPaletteSectionsSeparatedBySpacers(t *testing.T) {
+	// Section-named dashed headers are gone; groups are visually
+	// separated by a single non-selectable blank row. Verify that the
+	// build emits one such spacer between every pair of adjacent groups
+	// and never a leading spacer.
+	c := makeFakeChild("a", "claude", KindAgent)
+	other := makeFakeChild("b", "worker", KindCommand)
+	p := newPalette([]*Child{c, other}, "a", "",
+		preset.Set{Agents: []*preset.Preset{{Name: "codex"}}})
+
+	if len(p.items) == 0 {
+		t.Fatalf("palette built no items")
+	}
+	if p.items[0].action.kind == "header" {
+		t.Fatalf("first row is a spacer; should be a selectable item")
+	}
+	transitions := 0
+	prevGroup := p.items[0].group
+	for i := 1; i < len(p.items); i++ {
+		it := p.items[i]
+		if it.group != prevGroup {
+			if it.action.kind != "header" || it.label != "" {
+				t.Fatalf("group transition at %d not a blank spacer: %+v", i, it)
+			}
+			transitions++
+			// The row immediately after the spacer must be selectable.
+			if i+1 >= len(p.items) || p.items[i+1].action.kind == "header" {
+				t.Fatalf("spacer at %d not followed by selectable row", i)
+			}
+			prevGroup = p.items[i+1].group
+		}
+		// No dashed banners anywhere.
+		if it.action.kind == "header" && strings.Contains(it.label, "──") {
+			t.Errorf("dashed section header still present at %d: %q", i, it.label)
+		}
+	}
+	if transitions == 0 {
+		t.Fatalf("no section transitions found in palette items")
+	}
+}
+
+func TestPaletteCursorSkipsHeaders(t *testing.T) {
+	pr := []*preset.Preset{{Name: "a"}, {Name: "b"}}
+	p := newPalette(nil, "", "", preset.Set{Agents: pr})
+	// Initial cursor must land on a selectable row, never a header.
+	if p.items[p.cursor].action.kind == "header" {
+		t.Fatalf("initial cursor sits on a header: %+v", p.items[p.cursor])
+	}
+	// Walk to the end with cursorDown; every stop must be selectable.
+	for i := 0; i < len(p.items)*2; i++ {
+		p.cursorDown()
+		if p.items[p.cursor].action.kind == "header" {
+			t.Fatalf("cursorDown landed on a header at index %d", p.cursor)
+		}
+	}
+	// Walk back to top.
+	for i := 0; i < len(p.items)*2; i++ {
+		p.cursorUp()
+		if p.items[p.cursor].action.kind == "header" {
+			t.Fatalf("cursorUp landed on a header at index %d", p.cursor)
+		}
+	}
+}
+
+func TestPaletteEnterOnHeaderIsNoOp(t *testing.T) {
+	pr := []*preset.Preset{{Name: "a"}}
+	p := newPalette(nil, "", "", preset.Set{Agents: pr})
+	// Force the cursor onto a header.
+	for i, it := range p.items {
+		if it.action.kind == "header" {
+			p.cursor = i
+			break
+		}
+	}
+	_, done, _ := p.handleInput([]byte("\r"), 0)
+	if done {
+		t.Fatalf("Enter on header closed palette; expected no-op")
+	}
+}
+
+// -- Phase 3: filter chips & macro coexistence -----------------------
+
+func TestPaletteTabCyclesChip(t *testing.T) {
+	p := newTestPalette()
+	// All → Open
+	_, _, _ = p.handleInput([]byte{'\t'}, 0)
+	if string(p.query) != "sw " {
+		t.Fatalf("Tab #1: query %q, want %q", string(p.query), "sw ")
+	}
+	// Open → Spawn
+	_, _, _ = p.handleInput([]byte{'\t'}, 0)
+	if string(p.query) != "sp " {
+		t.Fatalf("Tab #2: query %q, want %q", string(p.query), "sp ")
+	}
+	// Spawn → Close
+	_, _, _ = p.handleInput([]byte{'\t'}, 0)
+	if string(p.query) != "k " {
+		t.Fatalf("Tab #3: query %q, want %q", string(p.query), "k ")
+	}
+	// Close → All (wraps)
+	_, _, _ = p.handleInput([]byte{'\t'}, 0)
+	if string(p.query) != "" {
+		t.Fatalf("Tab #4 wrap: query %q, want empty", string(p.query))
+	}
+}
+
+func TestPaletteShiftTabCyclesBackwards(t *testing.T) {
+	p := newTestPalette()
+	// Shift-Tab via legacy CSI Z: All → Close
+	_, _, _ = p.handleInput([]byte("\x1b[Z"), 0)
+	if string(p.query) != "k " {
+		t.Fatalf("Shift-Tab: query %q, want %q", string(p.query), "k ")
+	}
+}
+
+func TestPaletteBackspaceThroughTrailingMacro(t *testing.T) {
+	p := newTestPalette()
+	p.query = []rune("sw ")
+	p.rebuild()
+	p.backspace()
+	if string(p.query) != "" {
+		t.Fatalf("backspace through 'sw ' left %q; want empty", string(p.query))
+	}
+}
+
+func TestPaletteMacroPreservesQueryCase(t *testing.T) {
+	// Tab cycling shouldn't downcase the user-typed search text.
+	p := newTestPalette()
+	p.query = []rune("Foo")
+	p.rebuild()
+	_, _, _ = p.handleInput([]byte{'\t'}, 0)
+	if string(p.query) != "sw Foo" {
+		t.Fatalf("query after Tab over 'Foo' = %q; want 'sw Foo'", string(p.query))
+	}
+}
+
+// -- Phase 4: scored matching ----------------------------------------
+
+func TestFuzzyScorePrefixBeatsBoundaryBeatsSubstring(t *testing.T) {
+	prefix, _ := fuzzyScore("spawn agent: foo", "", "spa")
+	boundary, _ := fuzzyScore("hello spam", "", "spa")
+	substring, _ := fuzzyScore("escapade", "", "spa")
+	if !(prefix > boundary && boundary > substring) {
+		t.Fatalf("score ordering wrong: prefix=%d boundary=%d substring=%d", prefix, boundary, substring)
+	}
+}
+
+func TestFuzzyScoreReturnsMatchPositions(t *testing.T) {
+	_, pos := fuzzyScore("spawn process: dev", "", "dev")
+	want := []int{15, 16, 17}
+	if len(pos) != len(want) {
+		t.Fatalf("positions = %v, want %v", pos, want)
+	}
+	for i, p := range pos {
+		if p != want[i] {
+			t.Fatalf("pos[%d] = %d, want %d (full %v)", i, p, want[i], pos)
+		}
+	}
+}
+
+func TestPaletteScoredResultsDropHeaders(t *testing.T) {
+	pr := []*preset.Preset{{Name: "claude"}, {Name: "codex"}}
+	p := newPalette(nil, "", "", preset.Set{Agents: pr})
+	// Type a needle that matches both.
+	p.query = []rune("c")
+	p.rebuild()
+	for _, it := range p.items {
+		if it.action.kind == "header" {
+			t.Fatalf("scored mode should not emit header rows; got %+v", it)
+		}
+	}
+}
+
+func TestPaletteScoringFloatsPrefixMatchToTop(t *testing.T) {
+	// "x" is a prefix of "xtest" preset; it's a scattered-fuzzy match
+	// against many other rows. Scoring should land the prefix match at
+	// the top regardless of group order.
+	pr := []*preset.Preset{
+		{Name: "alpha"},
+		{Name: "xtest"},
+		{Name: "beta"},
+	}
+	p := newPalette(nil, "", "", preset.Set{Agents: pr})
+	p.query = []rune("xt")
+	p.rebuild()
+	if len(p.items) == 0 {
+		t.Fatalf("no scored items for needle 'xt'")
+	}
+	if !strings.Contains(p.items[0].label, "xtest") {
+		t.Fatalf("expected xtest at top of scored list, got %q", p.items[0].label)
+	}
+}
+
+// -- Phase 5: power-user accelerators --------------------------------
+
+func TestPaletteCtrlXOnSwitchKills(t *testing.T) {
+	c := makeFakeChild("a", "claude", KindAgent)
+	p := newPalette([]*Child{c}, "", "", preset.Set{})
+	// Cursor should already be on the switch row (it's the first
+	// selectable item with no Focused section).
+	idx, _ := findItem(p, "switch")
+	if idx < 0 {
+		t.Fatalf("no switch item in palette")
+	}
+	p.cursor = idx
+	action, done, _ := p.handleInput([]byte{0x18}, 0)
+	if !done {
+		t.Fatalf("Ctrl-X on switch row didn't close palette: action=%+v", action)
+	}
+	if action.kind != "kill" || action.childID != "a" {
+		t.Fatalf("Ctrl-X action = %+v, want kill of 'a'", action)
+	}
+}
+
+func TestPaletteCtrlXOnNonSwitchIsNoOp(t *testing.T) {
+	p := newPalette(nil, "", "", preset.Set{})
+	// Cursor parks on Quit or Spawn entries — neither is a switch row.
+	_, done, _ := p.handleInput([]byte{0x18}, 0)
+	if done {
+		t.Fatalf("Ctrl-X on non-switch closed palette")
+	}
+}
+
+func TestPaletteHelpToggle(t *testing.T) {
+	p := newTestPalette()
+	// `?` with empty query opens help.
+	_, done, _ := p.handleInput([]byte("?"), 0)
+	if done {
+		t.Fatalf("? closed palette")
+	}
+	if !p.showHelp {
+		t.Fatalf("? didn't open help")
+	}
+	// Next keystroke dismisses.
+	_, _, _ = p.handleInput([]byte("a"), 0)
+	if p.showHelp {
+		t.Fatalf("help still showing after dismissing keystroke")
+	}
+}
+
+func TestPaletteHelpDoesNotInterceptInQuery(t *testing.T) {
+	p := newTestPalette()
+	p.query = []rune("dev")
+	p.rebuild()
+	_, _, _ = p.handleInput([]byte("?"), 0)
+	if p.showHelp {
+		t.Fatalf("? with non-empty query incorrectly opened help")
+	}
+	if string(p.query) != "dev?" {
+		t.Fatalf("? with non-empty query failed to append: %q", string(p.query))
+	}
+}
+
+func TestPaletteHomeEndJumpsOverHeaders(t *testing.T) {
+	pr := []*preset.Preset{{Name: "a"}, {Name: "b"}}
+	p := newPalette(nil, "", "", preset.Set{Agents: pr})
+	// End jumps to last selectable.
+	p.cursorEnd()
+	if p.items[p.cursor].action.kind == "header" {
+		t.Fatalf("End landed on header: %+v", p.items[p.cursor])
+	}
+	if p.items[p.cursor].action.kind != "quit" {
+		t.Fatalf("End on simple palette should park on Quit; got %+v", p.items[p.cursor])
+	}
+	// Home returns to first selectable.
+	p.cursorHome()
+	if p.items[p.cursor].action.kind == "header" {
+		t.Fatalf("Home landed on header: %+v", p.items[p.cursor])
+	}
+}
+
+func TestPaletteAltDigitQuickPick(t *testing.T) {
+	pr := []*preset.Preset{{Name: "first"}, {Name: "second"}}
+	p := newPalette(nil, "", "", preset.Set{Agents: pr})
+	// Alt-1 picks the first selectable item (Spawn agent: first).
+	action, done, adv := p.handleInput([]byte("\x1b1"), 0)
+	if adv != 2 {
+		t.Fatalf("Alt-1 advance %d, want 2", adv)
+	}
+	if !done {
+		t.Fatalf("Alt-1 didn't close palette")
+	}
+	if action.kind != "spawn-agent" || action.preset == nil || action.preset.Name != "first" {
+		t.Fatalf("Alt-1 action = %+v, want spawn-agent first", action)
+	}
+}
+
+func TestAutoSummaryCadenceCyclesSoloValues(t *testing.T) {
+	p := newPalette(nil, "", "", preset.Set{}, defaultSettings())
+	p.mode = paletteModeAutoSummary
+	for i, row := range autoSummaryRows() {
+		if row.key == "cadence" {
+			p.cursor = i
+			break
+		}
+	}
+	if p.settings.AutoSummary.Cadence != "1m" {
+		t.Fatalf("initial cadence = %q", p.settings.AutoSummary.Cadence)
+	}
+	action, done, _ := p.activateAutoSummaryRow()
+	if done || action.kind != "settings-save" {
+		t.Fatalf("first cycle action = %+v done=%v, want settings-save without close", action, done)
+	}
+	if p.settings.AutoSummary.Cadence != "15s" {
+		t.Fatalf("first cycle cadence = %q", p.settings.AutoSummary.Cadence)
+	}
+	action, done, _ = p.activateAutoSummaryRow()
+	if done || action.kind != "settings-save" {
+		t.Fatalf("second cycle action = %+v done=%v, want settings-save without close", action, done)
+	}
+	if p.settings.AutoSummary.Cadence != "30s" {
+		t.Fatalf("second cycle cadence = %q", p.settings.AutoSummary.Cadence)
+	}
+	action, done, _ = p.activateAutoSummaryRow()
+	if done || action.kind != "settings-save" {
+		t.Fatalf("third cycle action = %+v done=%v, want settings-save without close", action, done)
+	}
+	if p.settings.AutoSummary.Cadence != "1m" {
+		t.Fatalf("third cycle cadence = %q", p.settings.AutoSummary.Cadence)
+	}
+}
+
+func TestAutoSummaryScreenOmitsExplicitSaveCancelBackRows(t *testing.T) {
+	omitted := map[string]bool{
+		"Save settings":    true,
+		"Cancel":           true,
+		"Back to Settings": true,
+	}
+	for _, row := range autoSummaryRows() {
+		if omitted[row.label] {
+			t.Fatalf("auto-summary settings should not show %q", row.label)
+		}
+	}
+}
+
+func TestAutoSummaryRenderOmitsStaleSettingsHelp(t *testing.T) {
+	p := newPalette(nil, "", "", preset.Set{}, defaultSettings())
+	p.mode = paletteModeAutoSummary
+	var b bytes.Buffer
+	p.renderAutoSummary(wrapWriter(&b), 100, 30)
+	out := b.String()
+	for _, text := range []string{
+		"Save settings",
+		"Cancel",
+		"Back to Settings",
+		"changes save",
+		"applies immediately",
+	} {
+		if strings.Contains(out, text) {
+			t.Fatalf("auto-summary render should not contain %q:\n%s", text, out)
+		}
+	}
+}
+
+func TestAutoSummaryValueRowsStyleLabelAndValueSeparately(t *testing.T) {
+	p := newPalette(nil, "", "", preset.Set{}, defaultSettings())
+	rows := p.autoSummaryDisplayRows()
+	for _, row := range rows {
+		if strings.Contains(row, "Cadence:") {
+			if !strings.HasPrefix(row, styleHint+"Cadence:"+styleReset+" ") {
+				t.Fatalf("cadence row styling = %q", row)
+			}
+			if strings.Contains(strings.TrimPrefix(row, styleHint+"Cadence:"+styleReset+" "), styleHint) {
+				t.Fatalf("cadence value should use regular text styling: %q", row)
+			}
+			return
+		}
+	}
+	t.Fatal("missing cadence display row")
+}
+
+func TestAutoSummaryTextInputSavesWhenSubmitted(t *testing.T) {
+	p := newPalette(nil, "", "", preset.Set{}, defaultSettings())
+	p.mode = paletteModeSettingsInput
+	p.settingsInput = &settingsInputForm{
+		title: "codex model",
+		field: "codex_model",
+		value: []rune("custom-model"),
+	}
+	action, done, _ := p.handleSettingsTextInput([]byte{'\r'}, 0)
+	if done || action.kind != "settings-save" {
+		t.Fatalf("submit action = %+v done=%v, want settings-save without close", action, done)
+	}
+	if got := p.settings.AutoSummary.modelFor("codex"); got != "custom-model" {
+		t.Fatalf("codex model = %q", got)
+	}
+}
+
+func TestPaletteFormCtrlRTogglesRelaunchFromCommandField(t *testing.T) {
+	p := newPalette(nil, "", "", preset.Set{})
+	p.mode = paletteModeSpawnForm
+	p.form = &spawnProcessForm{}
+	// Type without leaving the command field, then Ctrl-R.
+	for _, b := range []byte("xyz") {
+		_, _, _ = p.handleInput([]byte{b}, 0)
+	}
+	if p.form.field != 0 {
+		t.Fatalf("field jumped to %d", p.form.field)
+	}
+	_, _, _ = p.handleInput([]byte{0x12}, 0)
+	if !p.form.relaunch {
+		t.Fatalf("Ctrl-R didn't toggle relaunch from command field")
+	}
+	// Second press toggles back.
+	_, _, _ = p.handleInput([]byte{0x12}, 0)
+	if p.form.relaunch {
+		t.Fatalf("second Ctrl-R didn't toggle off")
+	}
+}
+
+// -- Phase 6: counter / scroll indicator -----------------------------
+
+func TestPaletteFooterCounter(t *testing.T) {
+	pr := []*preset.Preset{{Name: "a"}, {Name: "b"}, {Name: "c"}}
+	p := newPalette(nil, "", "", preset.Set{Agents: pr})
+	total := p.visibleSelectableCount()
+	if total < 4 { // 3 spawn-agents + terminal + custom + quit
+		t.Fatalf("expected ≥4 selectables; got %d", total)
+	}
+	idx := p.selectableIndex()
+	if idx <= 0 {
+		t.Fatalf("selectable index = %d on freshly-built palette; want ≥1", idx)
+	}
+}

internal/app/project_registry_test.go

@@ -0,0 +1,162 @@
+package app
+
+import (
+	"context"
+	"syscall"
+	"testing"
+
+	"github.com/hjbdev/patterm/internal/preset"
+)
+
+func TestSwitchProjectPreservesProjectProcessTrees(t *testing.T) {
+	t.Setenv("XDG_DATA_HOME", t.TempDir())
+	t.Setenv("XDG_CONFIG_HOME", t.TempDir())
+
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+
+	reg := newProjectRegistry(preset.Set{}, defaultSettings(), nil, 80, 24)
+	defer reg.Shutdown()
+
+	projectA, err := reg.Open(ctx, t.TempDir())
+	if err != nil {
+		t.Fatalf("open project A: %v", err)
+	}
+	projectB, err := reg.Open(ctx, t.TempDir())
+	if err != nil {
+		t.Fatalf("open project B: %v", err)
+	}
+
+	a, err := projectA.Session.Spawn(SpawnSpec{
+		Kind: KindCommand,
+		Argv: []string{"sh", "-c", "trap 'exit 0' TERM; while :; do sleep 1; done"},
+		Name: "a-loop",
+	}, 80, 24)
+	if err != nil {
+		t.Fatalf("spawn project A command: %v", err)
+	}
+	b, err := projectB.Session.Spawn(SpawnSpec{
+		Kind: KindCommand,
+		Argv: []string{"sh", "-c", "trap 'exit 0' TERM; while :; do sleep 1; done"},
+		Name: "b-loop",
+	}, 80, 24)
+	if err != nil {
+		t.Fatalf("spawn project B command: %v", err)
+	}
+	t.Cleanup(func() {
+		_ = projectA.Session.Kill(a.ID, syscall.SIGTERM)
+		_ = projectB.Session.Kill(b.ID, syscall.SIGTERM)
+	})
+	waitUntilLive(t, a)
+	waitUntilLive(t, b)
+
+	st := &uiState{
+		registry:   reg,
+		project:    projectA,
+		sess:       projectA.Session,
+		launcher:   projectA.Launcher,
+		pads:       projectA.Pads,
+		trust:      projectA.Trust,
+		timers:     projectA.Host.timers,
+		chromeWake: make(chan struct{}, 1),
+		view: ClientView{
+			ID:          "test",
+			ProjectKey:  projectA.Key,
+			ProjectName: projectA.Name,
+			Cols:        80,
+			Rows:        24,
+		},
+	}
+	st.focusChildLocked(a)
+	projectA.Session.Subscribe(st)
+
+	st.switchProject(projectB)
+	if st.view.ProjectKey != projectB.Key {
+		t.Fatalf("view project key = %q, want %q", st.view.ProjectKey, projectB.Key)
+	}
+	if st.sess != projectB.Session {
+		t.Fatalf("ui session did not move to project B")
+	}
+	if projectA.Session.FindChild(a.ID) == nil {
+		t.Fatalf("project A child disappeared after switch")
+	}
+	if projectB.Session.FindChild(b.ID) == nil {
+		t.Fatalf("project B child disappeared after switch")
+	}
+	if !a.IsLive() {
+		t.Fatalf("project A child stopped after switch")
+	}
+	if !b.IsLive() {
+		t.Fatalf("project B child stopped after switch")
+	}
+
+	st.switchProject(projectA)
+	if st.view.ProjectKey != projectA.Key {
+		t.Fatalf("view project key after switching back = %q, want %q", st.view.ProjectKey, projectA.Key)
+	}
+	if projectA.Session.FindChild(a.ID) == nil || projectB.Session.FindChild(b.ID) == nil {
+		t.Fatalf("switching back should preserve both project process trees")
+	}
+}
+
+func TestProjectRegistryScratchpadsRouteByCallerProject(t *testing.T) {
+	t.Setenv("XDG_DATA_HOME", t.TempDir())
+	t.Setenv("XDG_CONFIG_HOME", t.TempDir())
+
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+
+	reg := newProjectRegistry(preset.Set{}, defaultSettings(), nil, 80, 24)
+	defer reg.Shutdown()
+
+	projectA, err := reg.Open(ctx, t.TempDir())
+	if err != nil {
+		t.Fatalf("open project A: %v", err)
+	}
+	projectB, err := reg.Open(ctx, t.TempDir())
+	if err != nil {
+		t.Fatalf("open project B: %v", err)
+	}
+
+	a, err := projectA.Session.Spawn(SpawnSpec{
+		Kind: KindCommand,
+		Argv: []string{"sh", "-c", "trap 'exit 0' TERM; while :; do sleep 1; done"},
+		Name: "a-caller",
+	}, 80, 24)
+	if err != nil {
+		t.Fatalf("spawn project A caller: %v", err)
+	}
+	b, err := projectB.Session.Spawn(SpawnSpec{
+		Kind: KindCommand,
+		Argv: []string{"sh", "-c", "trap 'exit 0' TERM; while :; do sleep 1; done"},
+		Name: "b-caller",
+	}, 80, 24)
+	if err != nil {
+		t.Fatalf("spawn project B caller: %v", err)
+	}
+	t.Cleanup(func() {
+		_ = projectA.Session.Kill(a.ID, syscall.SIGTERM)
+		_ = projectB.Session.Kill(b.ID, syscall.SIGTERM)
+	})
+	waitUntilLive(t, a)
+	waitUntilLive(t, b)
+
+	if _, err := reg.ScratchpadWrite(a.ID, "note.md", "project A", ""); err != nil {
+		t.Fatalf("write project A scratchpad: %v", err)
+	}
+	if _, err := reg.ScratchpadWrite(b.ID, "note.md", "project B", ""); err != nil {
+		t.Fatalf("write project B scratchpad: %v", err)
+	}
+
+	gotA, _, err := reg.ScratchpadRead(a.ID, "note.md")
+	if err != nil {
+		t.Fatalf("read project A scratchpad: %v", err)
+	}
+	gotB, _, err := reg.ScratchpadRead(b.ID, "note.md")
+	if err != nil {
+		t.Fatalf("read project B scratchpad: %v", err)
+	}
+	if gotA != "project A" || gotB != "project B" {
+		t.Fatalf("scratchpad routing leaked between projects: A=%q B=%q", gotA, gotB)
+	}
+}

internal/app/ring_test.go

@@ -0,0 +1,147 @@
+package app
+
+import (
+	"bytes"
+	"testing"
+)
+
+func newRingChild() *Child {
+	return newChildEntry("id", "name", KindCommand, nil, nil, "", "", "")
+}
+
+func TestRingShortWrite(t *testing.T) {
+	c := newRingChild()
+	c.recordWrite([]byte("hello"))
+	b, end := c.StreamRead(0)
+	if end != 5 {
+		t.Fatalf("end=%d want 5", end)
+	}
+	if string(b) != "hello" {
+		t.Fatalf("got %q want %q", b, "hello")
+	}
+	// Read past the head returns nil, same end.
+	b, end = c.StreamRead(5)
+	if end != 5 || b != nil {
+		t.Fatalf("re-read: end=%d b=%v", end, b)
+	}
+}
+
+func TestRingIncrementalRead(t *testing.T) {
+	c := newRingChild()
+	c.recordWrite([]byte("abc"))
+	c.recordWrite([]byte("def"))
+	b, end := c.StreamRead(3)
+	if end != 6 || string(b) != "def" {
+		t.Fatalf("got %q end=%d", b, end)
+	}
+}
+
+func TestRingWrapAround(t *testing.T) {
+	c := newRingChild()
+	// Write more than ringCap to force wrap. Use a pattern we can
+	// verify: bytes equal to (i mod 256).
+	total := ringCap + 1000
+	src := make([]byte, total)
+	for i := range src {
+		src[i] = byte(i)
+	}
+	// Write in pieces to exercise the wrap copy in recordWrite.
+	for i := 0; i < total; i += 7777 {
+		end := i + 7777
+		if end > total {
+			end = total
+		}
+		c.recordWrite(src[i:end])
+	}
+	// The freshest ringCap bytes should be readable.
+	b, head := c.StreamRead(0)
+	if head != int64(total) {
+		t.Fatalf("head=%d want %d", head, total)
+	}
+	if len(b) != ringCap {
+		t.Fatalf("len(b)=%d want %d", len(b), ringCap)
+	}
+	want := src[total-ringCap:]
+	if !bytes.Equal(b, want) {
+		t.Fatalf("ring contents diverge from source tail")
+	}
+}
+
+func TestRingChunkLargerThanCap(t *testing.T) {
+	c := newRingChild()
+	src := make([]byte, ringCap+500)
+	for i := range src {
+		src[i] = byte(i + 1)
+	}
+	c.recordWrite(src)
+	b, head := c.StreamRead(0)
+	if head != int64(len(src)) {
+		t.Fatalf("head=%d want %d", head, len(src))
+	}
+	if len(b) != ringCap {
+		t.Fatalf("len(b)=%d want %d", len(b), ringCap)
+	}
+	if !bytes.Equal(b, src[500:]) {
+		t.Fatalf("ring tail mismatch")
+	}
+}
+
+func TestStripANSIBytesEquivalence(t *testing.T) {
+	cases := []string{
+		"hello world",
+		"\x1b[31mred\x1b[0m text",
+		"line1\nline2\r\nline3",
+		"bell\x07ish",
+		"weird \x1bA escape",
+		"truncated \x1b[1;",
+		"",
+	}
+	for _, in := range cases {
+		want := stripANSI(in)
+		got := string(stripANSIBytes(nil, []byte(in)))
+		if got != want {
+			t.Errorf("stripANSIBytes(%q) = %q want %q", in, got, want)
+		}
+	}
+}
+
+func TestNormalizeGridText(t *testing.T) {
+	cases := []struct {
+		name string
+		in   string
+		want string
+	}{
+		{
+			name: "line endings",
+			in:   "one\r\ntwo\rthree",
+			want: "one\ntwo\nthree",
+		},
+		{
+			name: "trailing whitespace",
+			in:   "one   \ntwo\t\t\nthree",
+			want: "one\ntwo\nthree",
+		},
+		{
+			name: "collapse blank runs",
+			in:   "one\n\n\n two\n \n\t\nthree",
+			want: "one\n\n two\n\nthree",
+		},
+		{
+			name: "trim leading and trailing blanks",
+			in:   "\n \n\t\none\n\n",
+			want: "one",
+		},
+		{
+			name: "already clean",
+			in:   "one\n\ntwo\nthree",
+			want: "one\n\ntwo\nthree",
+		},
+	}
+	for _, tc := range cases {
+		t.Run(tc.name, func(t *testing.T) {
+			if got := normalizeGridText(tc.in); got != tc.want {
+				t.Fatalf("normalizeGridText(%q) = %q, want %q", tc.in, got, tc.want)
+			}
+		})
+	}
+}

internal/app/scratchpad_delete_test.go

@@ -0,0 +1,137 @@
+package app
+
+import (
+	"errors"
+	"io"
+	"os"
+	"testing"
+
+	"github.com/hjbdev/patterm/internal/preset"
+	"github.com/hjbdev/patterm/internal/scratchpad"
+)
+
+func silenceStdout(t *testing.T) {
+	t.Helper()
+	old := os.Stdout
+	r, w, err := os.Pipe()
+	if err != nil {
+		t.Fatalf("pipe stdout: %v", err)
+	}
+	done := make(chan struct{})
+	go func() {
+		_, _ = io.Copy(io.Discard, r)
+		close(done)
+	}()
+	os.Stdout = w
+	t.Cleanup(func() {
+		os.Stdout = old
+		_ = w.Close()
+		<-done
+		_ = r.Close()
+	})
+}
+
+func newScratchpadDeleteTestState(t *testing.T) (*uiState, *scratchpad.Store) {
+	t.Helper()
+	t.Setenv("XDG_DATA_HOME", t.TempDir())
+	pads, err := scratchpad.Open("scratchpad-delete-test")
+	if err != nil {
+		t.Fatalf("scratchpad.Open: %v", err)
+	}
+	sess := NewSession(t.TempDir(), "scratchpad-delete-test")
+	t.Cleanup(sess.Shutdown)
+	st := &uiState{
+		sess:       sess,
+		pads:       pads,
+		hostCols:   120,
+		hostRows:   40,
+		chromeWake: make(chan struct{}, 1),
+	}
+	return st, pads
+}
+
+func TestDeletingFocusedScratchpadFocusesAnotherPad(t *testing.T) {
+	silenceStdout(t)
+	st, pads := newScratchpadDeleteTestState(t)
+	if _, err := pads.Write("alpha.md", "alpha", ""); err != nil {
+		t.Fatalf("write alpha: %v", err)
+	}
+	if _, err := pads.Write("beta.md", "beta", ""); err != nil {
+		t.Fatalf("write beta: %v", err)
+	}
+	st.focusedPad = "alpha.md"
+	st.focusedName = "alpha.md"
+	st.padOffsetName = "alpha.md"
+	st.padOffset = 3
+
+	st.handlePadDelete("alpha.md")
+
+	if st.focusedPad != "beta.md" {
+		t.Fatalf("focusedPad = %q, want beta.md", st.focusedPad)
+	}
+	if st.focusedID != "" {
+		t.Fatalf("focusedID = %q, want empty while another pad is focused", st.focusedID)
+	}
+	if st.padOffset != 0 || st.padOffsetName != "beta.md" {
+		t.Fatalf("pad offset = (%q,%d), want (beta.md,0)", st.padOffsetName, st.padOffset)
+	}
+}
+
+func TestDeletingLastFocusedScratchpadFocusesRunningChild(t *testing.T) {
+	silenceStdout(t)
+	st, pads := newScratchpadDeleteTestState(t)
+	if _, err := pads.Write("only.md", "only", ""); err != nil {
+		t.Fatalf("write only: %v", err)
+	}
+	child := makeFakeChild("pid", "devserver", KindCommand)
+	addChild(st.sess, child)
+	st.focusedPad = "only.md"
+	st.focusedName = "only.md"
+
+	st.handlePadDelete("only.md")
+
+	if st.focusedPad != "" {
+		t.Fatalf("focusedPad = %q, want empty after falling back to child", st.focusedPad)
+	}
+	if st.focusedID != "pid" {
+		t.Fatalf("focusedID = %q, want pid", st.focusedID)
+	}
+}
+
+type scratchpadChangeRecorder struct {
+	count int
+}
+
+func (r *scratchpadChangeRecorder) scratchpadsChanged() {
+	r.count++
+}
+
+func TestToolHostScratchpadDeleteRemovesPadAndRefreshes(t *testing.T) {
+	t.Setenv("XDG_DATA_HOME", t.TempDir())
+	pads, err := scratchpad.Open("scratchpad-delete-host-test")
+	if err != nil {
+		t.Fatalf("scratchpad.Open: %v", err)
+	}
+	if _, err := pads.Write("doomed.md", "content", ""); err != nil {
+		t.Fatalf("write doomed.md: %v", err)
+	}
+	recorder := &scratchpadChangeRecorder{}
+	host := newToolHost(nil, pads, nil, preset.Set{}, nil, 120, 40)
+	host.scratch = recorder
+
+	if err := host.ScratchpadDelete("", "doomed.md"); err != nil {
+		t.Fatalf("ScratchpadDelete: %v", err)
+	}
+	if recorder.count != 1 {
+		t.Fatalf("scratchpadsChanged calls = %d, want 1", recorder.count)
+	}
+	if _, _, err := pads.Read("doomed.md"); !errors.Is(err, os.ErrNotExist) {
+		t.Fatalf("read deleted pad error = %v, want os.ErrNotExist", err)
+	}
+	if err := host.ScratchpadDelete("", "doomed.md"); !errors.Is(err, os.ErrNotExist) {
+		t.Fatalf("delete missing error = %v, want os.ErrNotExist", err)
+	}
+	if recorder.count != 1 {
+		t.Fatalf("scratchpadsChanged calls after failed delete = %d, want 1", recorder.count)
+	}
+}

internal/app/session.go

@@ -12,9 +12,11 @@ import (
 	"fmt"
 	"os"
 	"sync"
+	"sync/atomic"
 	"syscall"
 	"time"
 
+	"github.com/hjbdev/patterm/internal/persist"
 	"github.com/hjbdev/patterm/internal/vt"
 )
 
@@ -38,8 +40,49 @@ type Session struct {
 
 	// listeners is the set of UI listeners that want to hear about child
 	// lifecycle events (spawn/exit) — exactly one (the TUI) in v1.
+	// listeners is an atomic.Pointer to a frozen slice. Subscribe
+	// copy-on-writes the slice; emit* paths use a single atomic Load.
+	// This drops one mutex acquisition per PTY chunk on the hot path.
 	listenersMu sync.Mutex
-	listeners   []ChildEventListener
+	listeners   atomic.Pointer[[]ChildEventListener]
+
+	// clientListeners is the network-client subscriber path. These
+	// listeners must be non-blocking and copy PTY chunks before enqueueing;
+	// daemon-internal observers (timers, debug capture, waiters) stay on
+	// listeners above so backpressure policy is isolated to clients.
+	clientListenersMu sync.Mutex
+	clientListeners   atomic.Pointer[[]ChildEventListener]
+
+	// persistStore records top-level command entries to a per-project
+	// JSON file so they can be re-spawned after patterm restarts.
+	// Optional; nil means "no persistence" (used by unit tests).
+	persistStore *persist.Store
+
+	// metrics is the optional performance tracker. nil when --profile
+	// is off. The pump goroutine reads it via atomic Load so installing
+	// metrics post-construction doesn't race with running children.
+	metrics atomic.Pointer[metricsTracker]
+}
+
+// SetPersistStore attaches a process-persistence store. Future Spawn /
+// Close / Rename / SetAutoRestart calls on top-level command entries
+// will mirror the change into the store.
+func (s *Session) SetPersistStore(p *persist.Store) {
+	s.mu.Lock()
+	s.persistStore = p
+	s.mu.Unlock()
+}
+
+// SetMetrics installs the per-session performance tracker. Safe to
+// call with nil to disable (the default). Reads on the hot path go
+// through atomic.Pointer.Load() with no lock; SetMetrics swaps the
+// pointer once at startup.
+func (s *Session) SetMetrics(m *metricsTracker) {
+	s.metrics.Store(m)
+}
+
+func (s *Session) loadMetrics() *metricsTracker {
+	return s.metrics.Load()
 }
 
 // ChildEventListener is implemented by the TUI to react to lifecycle
@@ -51,6 +94,16 @@ type ChildEventListener interface {
 	// Only the focused-child chunk should reach the screen — the TUI
 	// filters by id.
 	OnPTYOut(childID string, chunk []byte)
+	// OnChildStateChanged fires when the idle-detection classifier
+	// updates a child's IdleState. Listeners use this to repaint the
+	// sidebar badge and to evaluate idle-aware timers.
+	OnChildStateChanged(childID string, state IdleState)
+	// OnChildClosed fires when a child is being removed from the
+	// session (either via close_process, or — for agent/terminal
+	// kinds — when the PTY exits and the entry will never be
+	// restarted). It signals that any pending references to childID
+	// (e.g. timers owned by or watching it) should be dropped.
+	OnChildClosed(childID string)
 }
 
 func NewSession(projectDir, projectKey string) *Session {
@@ -65,34 +118,123 @@ func NewSession(projectDir, projectKey string) *Session {
 func (s *Session) Subscribe(l ChildEventListener) {
 	s.listenersMu.Lock()
 	defer s.listenersMu.Unlock()
-	s.listeners = append(s.listeners, l)
+	prev := s.listenersSnapshot()
+	next := make([]ChildEventListener, 0, len(prev)+1)
+	next = append(next, prev...)
+	next = append(next, l)
+	s.listeners.Store(&next)
+}
+
+func (s *Session) SubscribeClient(l ChildEventListener) {
+	s.clientListenersMu.Lock()
+	defer s.clientListenersMu.Unlock()
+	prev := s.clientListenersSnapshot()
+	next := make([]ChildEventListener, 0, len(prev)+1)
+	next = append(next, prev...)
+	next = append(next, l)
+	s.clientListeners.Store(&next)
+}
+
+// Unsubscribe removes a previously-registered listener. Safe to call
+// with a listener that wasn't registered (no-op).
+func (s *Session) Unsubscribe(l ChildEventListener) {
+	s.listenersMu.Lock()
+	defer s.listenersMu.Unlock()
+	prev := s.listenersSnapshot()
+	if len(prev) == 0 {
+		return
+	}
+	next := make([]ChildEventListener, 0, len(prev))
+	for _, e := range prev {
+		if e != l {
+			next = append(next, e)
+		}
+	}
+	s.listeners.Store(&next)
+}
+
+// UnsubscribeClient removes a previously-registered network client listener.
+// Safe to call with a listener that was never registered.
+func (s *Session) UnsubscribeClient(l ChildEventListener) {
+	s.clientListenersMu.Lock()
+	defer s.clientListenersMu.Unlock()
+	prev := s.clientListenersSnapshot()
+	if len(prev) == 0 {
+		return
+	}
+	next := make([]ChildEventListener, 0, len(prev))
+	for _, e := range prev {
+		if e != l {
+			next = append(next, e)
+		}
+	}
+	s.clientListeners.Store(&next)
+}
+
+// listenersSnapshot returns the frozen listener slice. Safe to call
+// without the listeners mutex.
+func (s *Session) listenersSnapshot() []ChildEventListener {
+	p := s.listeners.Load()
+	if p == nil {
+		return nil
+	}
+	return *p
+}
+
+func (s *Session) clientListenersSnapshot() []ChildEventListener {
+	p := s.clientListeners.Load()
+	if p == nil {
+		return nil
+	}
+	return *p
 }
 
 func (s *Session) emitSpawn(c *Child) {
-	s.listenersMu.Lock()
-	ls := append([]ChildEventListener(nil), s.listeners...)
-	s.listenersMu.Unlock()
-	for _, l := range ls {
+	for _, l := range s.listenersSnapshot() {
+		l.OnChildSpawned(c)
+	}
+	for _, l := range s.clientListenersSnapshot() {
 		l.OnChildSpawned(c)
 	}
 }
 
 func (s *Session) emitExit(c *Child) {
-	s.listenersMu.Lock()
-	ls := append([]ChildEventListener(nil), s.listeners...)
-	s.listenersMu.Unlock()
-	for _, l := range ls {
+	for _, l := range s.listenersSnapshot() {
+		l.OnChildExited(c)
+	}
+	for _, l := range s.clientListenersSnapshot() {
 		l.OnChildExited(c)
 	}
 }
 
+// emitPTYOut dispatches a fresh PTY chunk to every listener. Listeners
+// MUST NOT retain `chunk` past return — the slice is owned by the
+// pumpChild read buffer and is overwritten on the next read.
 func (s *Session) emitPTYOut(id string, chunk []byte) {
-	s.listenersMu.Lock()
-	ls := append([]ChildEventListener(nil), s.listeners...)
-	s.listenersMu.Unlock()
-	for _, l := range ls {
+	for _, l := range s.listenersSnapshot() {
 		l.OnPTYOut(id, chunk)
 	}
+	for _, l := range s.clientListenersSnapshot() {
+		l.OnPTYOut(id, chunk)
+	}
+}
+
+func (s *Session) emitStateChanged(id string, state IdleState) {
+	for _, l := range s.listenersSnapshot() {
+		l.OnChildStateChanged(id, state)
+	}
+	for _, l := range s.clientListenersSnapshot() {
+		l.OnChildStateChanged(id, state)
+	}
+}
+
+func (s *Session) emitClosed(id string) {
+	for _, l := range s.listenersSnapshot() {
+		l.OnChildClosed(id)
+	}
+	for _, l := range s.clientListenersSnapshot() {
+		l.OnChildClosed(id)
+	}
 }
 
 func (s *Session) ChildEnv() []string {
@@ -123,6 +265,11 @@ type SpawnSpec struct {
 	// or is closed. They must be attached before the PTY starts so a
 	// fast-exiting child cannot outrun cleanup registration.
 	CleanupPaths []string
+	// IdleDetection is the resolved per-preset idle classifier config.
+	// Must be installed before the child is published to s.children so
+	// the classifier goroutine never observes a nil/default config for
+	// a preset that overrides it.
+	IdleDetection *resolvedIdleDetection
 }
 
 // Spawn creates a new entry and starts its PTY. For Kind = command the
@@ -137,6 +284,9 @@ func (s *Session) Spawn(spec SpawnSpec, cols, rows uint16) (*Child, error) {
 	if spec.Env == nil {
 		spec.Env = s.ChildEnv()
 	}
+	if spec.WorkDir == "" {
+		spec.WorkDir = s.projectDir
+	}
 
 	s.mu.Lock()
 	id := s.mintUniqueIDLocked()
@@ -153,6 +303,12 @@ func (s *Session) Spawn(spec SpawnSpec, cols, rows uint16) (*Child, error) {
 	for _, path := range spec.CleanupPaths {
 		c.AddCleanupPath(path)
 	}
+	// Install idle-detection BEFORE publishing to s.children — otherwise
+	// the classifier goroutine could read c.idleDetection while the
+	// launcher is still racing to set it.
+	if spec.IdleDetection != nil {
+		c.setIdleDetection(spec.IdleDetection)
+	}
 	runID, err := c.startPTY(cols, rows)
 	if err != nil {
 		c.cleanupOwnedPaths()
@@ -162,14 +318,67 @@ func (s *Session) Spawn(spec SpawnSpec, cols, rows uint16) (*Child, error) {
 	s.mu.Lock()
 	s.children[id] = c
 	s.order = append(s.order, id)
+	store := s.persistStore
 	s.mu.Unlock()
 
+	// Wire persistence callback BEFORE registering so SetName /
+	// SetAutoRestart calls that race the listener still hit the store.
+	if store != nil {
+		c.setPersistFn(func(ch *Child) {
+			s.persistEntry(ch)
+		})
+		s.persistEntry(c)
+	}
+
 	s.emitSpawn(c)
 	go s.pumpChild(c, runID)
 	go s.reapChild(c, runID)
 	return c, nil
 }
 
+// persistEntry writes (or refreshes) the persist record for c if it
+// qualifies — top-level command entries only. No-op when no store is
+// attached.
+func (s *Session) persistEntry(c *Child) {
+	s.mu.Lock()
+	store := s.persistStore
+	s.mu.Unlock()
+	if store == nil || !shouldPersist(c) {
+		return
+	}
+	e := persist.Entry{
+		ID:          c.ID,
+		Name:        c.DisplayName(),
+		Argv:        append([]string(nil), c.Argv...),
+		WorkDir:     c.WorkDir,
+		PresetRef:   c.PresetRef,
+		AutoRestart: c.AutoRestart(),
+	}
+	if err := store.Save(e); err != nil {
+		logf("persist save %s: %v", c.ID, err)
+	}
+}
+
+func (s *Session) forgetPersisted(id string) {
+	s.mu.Lock()
+	store := s.persistStore
+	s.mu.Unlock()
+	if store == nil {
+		return
+	}
+	if err := store.Remove(id); err != nil {
+		logf("persist remove %s: %v", id, err)
+	}
+}
+
+// shouldPersist gates which Child entries get mirrored into the
+// persist store. v1 only restores top-level command entries — agents
+// and terminals are ephemeral by design, and sub-agent-spawned
+// commands belong to their orchestrator's lifecycle.
+func shouldPersist(c *Child) bool {
+	return c != nil && c.Kind == KindCommand && c.ParentID == ""
+}
+
 // Start (re)attaches a PTY to an entry that is currently stopped or
 // exited. Errors if the entry is already live.
 func (s *Session) Start(id string, cols, rows uint16) error {
@@ -238,6 +447,26 @@ func (s *Session) Close(id string, sig syscall.Signal) error {
 		}
 	}
 	s.mu.Unlock()
+	// Notify listeners outside s.mu so they can take their own locks
+	// without inversion. Timer manager uses this to drop pending
+	// timers owned by or watching the closed child — otherwise the
+	// next classifier tick can deliver a stale fire to the parent.
+	s.emitClosed(id)
+	s.forgetPersisted(id)
+	return nil
+}
+
+// Terminate stops a live child with SIGTERM/SIGKILL escalation but
+// leaves its session entry intact so callers can keep showing the
+// exited pane.
+func (s *Session) Terminate(id string, sig syscall.Signal) error {
+	c := s.FindChild(id)
+	if c == nil {
+		return fmt.Errorf("no such process %q", id)
+	}
+	if c.IsLive() {
+		terminateAndWait(c, sig, childStopTimeout)
+	}
 	return nil
 }
 
@@ -257,6 +486,12 @@ func (s *Session) pumpChild(c *Child, runID uint64) {
 	if pty == nil {
 		return
 	}
+	// One PTY read buffer per pump goroutine. Consumers downstream
+	// (em.Write is synchronous through CGO; recordWrite append-copies
+	// into the ring; renderer.Render copies into its pending buffer)
+	// all complete or copy before returning, so the buffer can be
+	// reused without aliasing live data. See ChildEventListener.OnPTYOut
+	// docstring — listeners must not retain `chunk`.
 	buf := make([]byte, 64*1024)
 	for {
 		n, err := pty.Read(buf)
@@ -264,12 +499,40 @@ func (s *Session) pumpChild(c *Child, runID uint64) {
 			if !c.isCurrentRun(runID) {
 				return
 			}
-			chunk := make([]byte, n)
-			copy(chunk, buf[:n])
+			chunk := buf[:n]
 			if em := c.Emulator(); em != nil {
+				m := s.loadMetrics()
+				wstart := time.Time{}
+				if m != nil {
+					wstart = time.Now()
+				}
 				if _, werr := em.Write(chunk); werr != nil {
 					logf("emulator.Write(child %s): %v", c.ID, werr)
 				}
+				if m != nil {
+					m.recordEmuWrite(time.Since(wstart))
+				}
+				// OSC 0/2 title updates ride on the same byte stream as
+				// the rest of the output. Polling the emulator after each
+				// chunk is cheap on its own (one CGO call) but codex/
+				// ratatui sends so many small chunks that the per-chunk
+				// CGO cost becomes measurable. Skip the Title poll when
+				// the chunk doesn't carry an OSC start byte at all; the
+				// title can only change on chunks that include one.
+				if containsOSC(chunk) {
+					tstart := time.Time{}
+					if m != nil {
+						tstart = time.Now()
+					}
+					if t, terr := em.Title(); terr == nil {
+						c.recordTitle(t)
+					}
+					if m != nil {
+						m.recordEmuTitle(time.Since(tstart), false)
+					}
+				} else if m != nil {
+					m.recordEmuTitle(0, true)
+				}
 			}
 			c.recordWrite(chunk)
 			s.emitPTYOut(c.ID, chunk)
@@ -299,6 +562,24 @@ func (s *Session) reapChild(c *Child, runID uint64) {
 	if !c.restarting.Load() {
 		c.cleanupOwnedPaths()
 	}
+	// Terminals are ephemeral: unlike command entries (kept around for
+	// restart_process) and agents (which the user clears via close_process
+	// once they're done with the corpse), an exited terminal has nothing
+	// useful left to do. Drop it from the session so it disappears from
+	// the Processes sidebar / switch list immediately.
+	if c.Kind == KindTerminal && !c.restarting.Load() {
+		c.teardownPTY()
+		s.mu.Lock()
+		delete(s.children, c.ID)
+		for i, oid := range s.order {
+			if oid == c.ID {
+				s.order = append(s.order[:i], s.order[i+1:]...)
+				break
+			}
+		}
+		s.mu.Unlock()
+		s.emitClosed(c.ID)
+	}
 }
 
 // killDescendantsOf terminates every still-live direct child of
@@ -461,6 +742,22 @@ func (s *Session) ResizeAll(cols, rows uint16) {
 	}
 }
 
+func (s *Session) ResizeChild(id string, cols, rows uint16) {
+	if cols == 0 || rows == 0 {
+		return
+	}
+	c := s.FindChild(id)
+	if c == nil {
+		return
+	}
+	if pty := c.PTY(); pty != nil {
+		_ = pty.Resize(cols, rows)
+	}
+	if em := c.Emulator(); em != nil {
+		_ = em.Resize(cols, rows)
+	}
+}
+
 // SerializeChild returns the VT bytes that reproduce the child's
 // current screen state. Used to repaint a child after the user switches
 // focus or closes the palette.
@@ -528,6 +825,24 @@ func (s *Session) Shutdown() {
 	}
 }
 
+// containsOSC reports whether chunk holds a sequence that could begin
+// an OSC. OSC starts as ESC ] (0x1b 0x5d) or the bare C1 ] (0x9d),
+// so a chunk without either cannot have changed the emulator's OSC
+// title state. Used to short-circuit the per-chunk Title() poll from
+// pumpChild, which otherwise pays a CGO call for every chunk even
+// when codex/ratatui is just emitting SGR-styled output.
+func containsOSC(chunk []byte) bool {
+	for i, b := range chunk {
+		if b == 0x9d {
+			return true
+		}
+		if b == 0x1b && i+1 < len(chunk) && chunk[i+1] == ']' {
+			return true
+		}
+	}
+	return false
+}
+
 func logf(format string, args ...any) {
 	if os.Getenv("PATTERM_DEBUG_LOG") == "" {
 		return

internal/app/session_test.go

@@ -1,6 +1,7 @@
 package app
 
 import (
+	"strings"
 	"syscall"
 	"testing"
 	"time"
@@ -57,6 +58,94 @@ func TestParentExitKillsDescendants(t *testing.T) {
 	waitUntilNotLive(t, grandchild)
 }
 
+// TestSpawnInstallsIdleDetectionBeforePublish guarantees that a child
+// spawned with SpawnSpec.IdleDetection has its resolved config visible
+// the instant the child appears in s.children — closing the race where
+// the classifier could read c.idleDetection before the launcher set it.
+func TestSpawnInstallsIdleDetectionBeforePublish(t *testing.T) {
+	sess := NewSession(t.TempDir(), "test")
+	want := &resolvedIdleDetection{
+		strategy:        StrategyOSCTitleStability,
+		idleThresholdMS: 9999,
+	}
+	c, err := sess.Spawn(SpawnSpec{
+		Kind:          KindCommand,
+		Argv:          []string{"sh", "-c", "sleep 30"},
+		IdleDetection: want,
+	}, 80, 24)
+	if err != nil {
+		t.Fatalf("spawn: %v", err)
+	}
+	defer func() { _ = c.signal(syscall.SIGTERM) }()
+
+	// Read back via the same access path the classifier uses
+	// (sess.Children) so the test fails if the field is set only
+	// AFTER the child is published.
+	var found *Child
+	for _, ch := range sess.Children() {
+		if ch.ID == c.ID {
+			found = ch
+			break
+		}
+	}
+	if found == nil {
+		t.Fatalf("spawned child %s not in Children()", c.ID)
+	}
+	if found.idleDetection == nil {
+		t.Fatalf("idleDetection nil after Spawn returned")
+	}
+	if found.idleDetection.strategy != StrategyOSCTitleStability {
+		t.Fatalf("strategy: got %q want %q", found.idleDetection.strategy, StrategyOSCTitleStability)
+	}
+	if found.idleDetection.idleThresholdMS != 9999 {
+		t.Fatalf("threshold: got %d want 9999", found.idleDetection.idleThresholdMS)
+	}
+}
+
+func TestTerminateEscalatesWithoutRemovingEntry(t *testing.T) {
+	sess := NewSession(t.TempDir(), "test")
+	c, err := sess.Spawn(SpawnSpec{
+		Kind: KindAgent,
+		Argv: []string{"sh", "-c", "trap '' TERM; echo ready; while :; do sleep 1; done"},
+	}, 80, 24)
+	if err != nil {
+		t.Fatalf("spawn: %v", err)
+	}
+	t.Cleanup(func() {
+		if c.IsLive() {
+			_ = c.signal(syscall.SIGKILL)
+		}
+	})
+	waitUntilLive(t, c)
+	waitForStreamText(t, c, "ready")
+
+	start := time.Now()
+	if err := sess.Terminate(c.ID, syscall.SIGTERM); err != nil {
+		t.Fatalf("Terminate: %v", err)
+	}
+	if elapsed := time.Since(start); elapsed < childStopTimeout {
+		t.Fatalf("Terminate returned before SIGKILL fallback: elapsed=%s timeout=%s", elapsed, childStopTimeout)
+	}
+	waitUntilNotLive(t, c)
+
+	if got := sess.FindChild(c.ID); got == nil {
+		t.Fatalf("Terminate removed child entry %s", c.ID)
+	}
+}
+
+func waitForStreamText(t *testing.T, c *Child, want string) {
+	t.Helper()
+	deadline := time.Now().Add(5 * time.Second)
+	for time.Now().Before(deadline) {
+		b, _ := c.StreamRead(0)
+		if strings.Contains(string(b), want) {
+			return
+		}
+		time.Sleep(20 * time.Millisecond)
+	}
+	t.Fatalf("child %s never wrote %q", c.ID, want)
+}
+
 func waitUntilLive(t *testing.T, c *Child) {
 	t.Helper()
 	deadline := time.Now().Add(5 * time.Second)

internal/app/settings.go

@@ -0,0 +1,150 @@
+package app
+
+import (
+	"encoding/json"
+	"fmt"
+	"os"
+	"path/filepath"
+
+	"github.com/hjbdev/patterm/internal/preset"
+)
+
+const (
+	defaultSummaryProvider = "codex"
+	defaultCodexModel      = "gpt-5.4-mini"
+	defaultOpenCodeModel   = "opencode-go/minimax-m2.7"
+	defaultClaudeModel     = "claude-haiku-4-5"
+)
+
+type settings struct {
+	AutoSummary autoSummarySettings `json:"auto_summary"`
+}
+
+type autoSummarySettings struct {
+	Enabled         bool              `json:"enabled"`
+	Provider        string            `json:"provider"`
+	Models          map[string]string `json:"models"`
+	Cadence         string            `json:"cadence"`
+	QuietWindowMS   int               `json:"quiet_window_ms"`
+	MinInputChars   int               `json:"min_input_chars"`
+	MaxHistoryChars int               `json:"max_history_chars"`
+}
+
+func defaultSettings() settings {
+	return settings{
+		AutoSummary: autoSummarySettings{
+			Enabled:         true,
+			Provider:        defaultSummaryProvider,
+			Models:          defaultSummaryModels(),
+			Cadence:         "1m",
+			QuietWindowMS:   3000,
+			MinInputChars:   4,
+			MaxHistoryChars: 12000,
+		},
+	}
+}
+
+func defaultSummaryModels() map[string]string {
+	return map[string]string{
+		"codex":    defaultCodexModel,
+		"opencode": defaultOpenCodeModel,
+		"claude":   defaultClaudeModel,
+	}
+}
+
+func loadSettings() (settings, string, error) {
+	base, err := preset.ConfigDir()
+	if err != nil {
+		return settings{}, "", err
+	}
+	path := filepath.Join(base, "settings.json")
+	st := defaultSettings()
+	b, err := os.ReadFile(path)
+	if err != nil {
+		if os.IsNotExist(err) {
+			return st, path, nil
+		}
+		return st, path, fmt.Errorf("settings: read %s: %w", path, err)
+	}
+	if err := json.Unmarshal(b, &st); err != nil {
+		return defaultSettings(), path, fmt.Errorf("settings: parse %s: %w", path, err)
+	}
+	st.normalize()
+	return st, path, nil
+}
+
+func saveSettings(path string, st settings) error {
+	if path == "" {
+		return fmt.Errorf("settings: empty path")
+	}
+	st.normalize()
+	if err := os.MkdirAll(filepath.Dir(path), 0o700); err != nil {
+		return err
+	}
+	b, err := json.MarshalIndent(st, "", "  ")
+	if err != nil {
+		return err
+	}
+	b = append(b, '\n')
+	return os.WriteFile(path, b, 0o600)
+}
+
+func (st *settings) normalize() {
+	def := defaultSettings()
+	if st.AutoSummary.Provider == "" {
+		st.AutoSummary.Provider = def.AutoSummary.Provider
+	}
+	switch st.AutoSummary.Provider {
+	case "codex", "opencode", "claude":
+	default:
+		st.AutoSummary.Provider = def.AutoSummary.Provider
+	}
+	if st.AutoSummary.Models == nil {
+		st.AutoSummary.Models = defaultSummaryModels()
+	} else {
+		for k, v := range defaultSummaryModels() {
+			if st.AutoSummary.Models[k] == "" {
+				st.AutoSummary.Models[k] = v
+			}
+		}
+	}
+	if st.AutoSummary.Cadence == "" {
+		st.AutoSummary.Cadence = def.AutoSummary.Cadence
+	}
+	if st.AutoSummary.QuietWindowMS <= 0 {
+		st.AutoSummary.QuietWindowMS = def.AutoSummary.QuietWindowMS
+	}
+	if st.AutoSummary.MinInputChars <= 0 {
+		st.AutoSummary.MinInputChars = def.AutoSummary.MinInputChars
+	}
+	if st.AutoSummary.MaxHistoryChars <= 0 {
+		st.AutoSummary.MaxHistoryChars = def.AutoSummary.MaxHistoryChars
+	}
+}
+
+func (st settings) clone() settings {
+	st.normalize()
+	if st.AutoSummary.Models != nil {
+		models := make(map[string]string, len(st.AutoSummary.Models))
+		for k, v := range st.AutoSummary.Models {
+			models[k] = v
+		}
+		st.AutoSummary.Models = models
+	}
+	return st
+}
+
+func (a autoSummarySettings) clone() autoSummarySettings {
+	st := settings{AutoSummary: a}.clone()
+	return st.AutoSummary
+}
+
+func (a autoSummarySettings) modelFor(provider string) string {
+	if a.Models == nil {
+		return defaultSummaryModels()[provider]
+	}
+	if m := a.Models[provider]; m != "" {
+		return m
+	}
+	return defaultSummaryModels()[provider]
+}

internal/app/settings_test.go

@@ -0,0 +1,72 @@
+package app
+
+import (
+	"os"
+	"path/filepath"
+	"testing"
+)
+
+func TestLoadSettingsDefaults(t *testing.T) {
+	t.Setenv("XDG_CONFIG_HOME", t.TempDir())
+	st, path, err := loadSettings()
+	if err != nil {
+		t.Fatalf("loadSettings: %v", err)
+	}
+	if filepath.Base(path) != "settings.json" {
+		t.Fatalf("settings path = %q", path)
+	}
+	if !st.AutoSummary.Enabled {
+		t.Fatal("auto-summary should default enabled")
+	}
+	if st.AutoSummary.Provider != "codex" {
+		t.Fatalf("provider = %q want codex", st.AutoSummary.Provider)
+	}
+	if st.AutoSummary.Cadence != "1m" {
+		t.Fatalf("cadence = %q want 1m", st.AutoSummary.Cadence)
+	}
+	if got := st.AutoSummary.modelFor("codex"); got != "gpt-5.4-mini" {
+		t.Fatalf("codex model = %q", got)
+	}
+	if got := st.AutoSummary.modelFor("opencode"); got != "opencode-go/minimax-m2.7" {
+		t.Fatalf("opencode model = %q", got)
+	}
+}
+
+func TestSettingsCloneDoesNotShareModelMap(t *testing.T) {
+	st := defaultSettings()
+	cp := st.clone()
+	cp.AutoSummary.Models["codex"] = "changed"
+	if st.AutoSummary.Models["codex"] == "changed" {
+		t.Fatal("clone shared Models map with original")
+	}
+	a := st.AutoSummary.clone()
+	a.Models["opencode"] = "changed"
+	if st.AutoSummary.Models["opencode"] == "changed" {
+		t.Fatal("autoSummarySettings clone shared Models map with original")
+	}
+}
+
+func TestSaveAndLoadSettings(t *testing.T) {
+	dir := t.TempDir()
+	t.Setenv("XDG_CONFIG_HOME", dir)
+	st := defaultSettings()
+	st.AutoSummary.Provider = "opencode"
+	st.AutoSummary.Models["opencode"] = "minimax/test"
+	path := filepath.Join(dir, "patterm", "settings.json")
+	if err := saveSettings(path, st); err != nil {
+		t.Fatalf("saveSettings: %v", err)
+	}
+	if _, err := os.Stat(path); err != nil {
+		t.Fatalf("settings file missing: %v", err)
+	}
+	got, _, err := loadSettings()
+	if err != nil {
+		t.Fatalf("loadSettings: %v", err)
+	}
+	if got.AutoSummary.Provider != "opencode" {
+		t.Fatalf("provider = %q", got.AutoSummary.Provider)
+	}
+	if got.AutoSummary.modelFor("opencode") != "minimax/test" {
+		t.Fatalf("opencode model = %q", got.AutoSummary.modelFor("opencode"))
+	}
+}

internal/app/sidebar.go

@@ -4,6 +4,7 @@ import (
 	"fmt"
 	"os"
 	"strings"
+	"time"
 )
 
 const (
@@ -11,6 +12,146 @@ const (
 	statusRows  = 1
 )
 
+// fitName returns name truncated to fit budget visible cells, with a
+// trailing "…" when it overflows. Operates on RAW (unstyled) input;
+// the caller wraps the result in SGR. Returns "" when budget <= 0.
+func fitName(name string, budget int) string {
+	if budget <= 0 {
+		return ""
+	}
+	runes := []rune(name)
+	if len(runes) <= budget {
+		return name
+	}
+	if budget == 1 {
+		return "…"
+	}
+	return string(runes[:budget-1]) + "…"
+}
+
+// marqueeWindow returns the window of name starting at offset, exactly
+// budget cells wide. Pre: caller has decided the name overflows budget
+// and offset is in [0, len([]rune(name))-budget]. Operates on RAW
+// (unstyled) input.
+func marqueeWindow(name string, budget, offset int) string {
+	if budget <= 0 {
+		return ""
+	}
+	runes := []rune(name)
+	if len(runes) <= budget {
+		return name
+	}
+	if offset < 0 {
+		offset = 0
+	}
+	end := offset + budget
+	if end > len(runes) {
+		end = len(runes)
+		offset = end - budget
+		if offset < 0 {
+			offset = 0
+		}
+	}
+	return string(runes[offset:end])
+}
+
+// clampVisible truncates s so that its visible (non-SGR) length is at
+// most width cells, preserving any active style by appending a reset.
+// Used as a defensive net by write() so a row whose decoration was
+// mis-sized still cannot spill past the sidebar band into the PTY area.
+func clampVisible(s string, width int) string {
+	if width <= 0 {
+		return ""
+	}
+	if visibleLen(s) <= width {
+		return s
+	}
+	var b strings.Builder
+	b.Grow(len(s))
+	visible := 0
+	inEsc := false
+	for _, r := range s {
+		if inEsc {
+			b.WriteRune(r)
+			if r == 'm' || r == 'H' {
+				inEsc = false
+			}
+			continue
+		}
+		if r == 0x1b {
+			inEsc = true
+			b.WriteRune(r)
+			continue
+		}
+		if visible >= width {
+			break
+		}
+		b.WriteRune(r)
+		visible++
+	}
+	b.WriteString(styleReset)
+	return b.String()
+}
+
+// chooseSidebarSuffix decides whether to keep or drop the trailing
+// timer indicator from a sidebar row's suffix. When the row's name
+// would have to ellipsise with the timer present, but the budget
+// freed by dropping the timer still leaves at least 6 cells for the
+// name, the timer is dropped. The name is the only identifier the
+// user has for that row; the timer is recoverable from the status
+// line and palette.
+func chooseSidebarSuffix(nameRuneLen, width int, prefix, suffix, timer string) (string, int) {
+	prefixCost := visibleLen(prefix)
+	budget := width - prefixCost - visibleLen(suffix)
+	if nameRuneLen <= budget || timer == "" {
+		return suffix, budget
+	}
+	slim := strings.TrimSuffix(suffix, timer)
+	if slim == suffix {
+		return suffix, budget
+	}
+	slimBudget := width - prefixCost - visibleLen(slim)
+	if slimBudget >= 6 {
+		return slim, slimBudget
+	}
+	return suffix, budget
+}
+
+// rowNameSlot returns the unstyled name cell for a sidebar row.
+// Unfocused (or focused-and-fitting) rows get fitName with a trailing
+// "…" on overflow. The focused row, when its name overflows the
+// budget, gets the current marquee window — exactly budget cells
+// wide so the surrounding row geometry stays put while it animates.
+func (st *uiState) rowNameSlot(id, rawName string, budget int, focused bool) string {
+	if budget <= 0 {
+		return ""
+	}
+	runes := []rune(rawName)
+	if !focused || len(runes) <= budget {
+		return fitName(rawName, budget)
+	}
+	off, _, _ := st.marquee.step(id, len(runes), budget, time.Now())
+	return marqueeWindow(rawName, budget, off)
+}
+
+// formatShortDuration renders a duration as a short, sidebar-friendly
+// suffix: ms under 1s, "12s" under 60s, "3m" otherwise.
+func formatShortDuration(d time.Duration) string {
+	if d <= 0 {
+		return "0s"
+	}
+	if d < time.Second {
+		return fmt.Sprintf("%dms", int(d/time.Millisecond))
+	}
+	if d < time.Minute {
+		return fmt.Sprintf("%ds", int(d/time.Second))
+	}
+	if d < time.Hour {
+		return fmt.Sprintf("%dm", int(d/time.Minute))
+	}
+	return fmt.Sprintf("%dh", int(d/time.Hour))
+}
+
 // drawSidebar paints the right-rail session tree + scratchpad list.
 // SPEC §4: the rail is the active session's child hierarchy on top and
 // the scratchpad list (with preview) on the bottom.
@@ -19,9 +160,14 @@ const (
 // computed main viewport, so the sidebar region is outside the child's
 // cursor range. We can redraw freely without fighting the child for cells.
 func (st *uiState) drawSidebar() {
+	var entry time.Time
+	if st.metrics != nil {
+		entry = time.Now()
+	}
 	st.mu.Lock()
 	palOpen := st.palette != nil
 	focus := st.focusedID
+	focusPad := st.focusedPad
 	activeAgent := st.activeAgentID
 	st.mu.Unlock()
 	if palOpen {
@@ -49,6 +195,9 @@ func (st *uiState) drawSidebar() {
 		if row > maxRow {
 			return
 		}
+		if visibleLen(content) > width {
+			content = clampVisible(content, width)
+		}
 		pad := width - visibleLen(content)
 		if pad < 0 {
 			pad = 0
@@ -61,14 +210,56 @@ func (st *uiState) drawSidebar() {
 		write(" " + styleActive + text + styleReset)
 		write(" " + styleBorder + strings.Repeat("─", width-2) + styleReset)
 	}
+	// timerIndicator returns a short " ⏱ 12s" or " ⏸ paused" suffix
+	// when c has a pending or paused timer attached (owns or watches).
+	// Empty string when no timer is in play.
+	timerIndicator := func(c *Child) string {
+		if st.timers == nil {
+			return ""
+		}
+		info := st.timers.activeForChild(c.ID)
+		if info == nil {
+			return ""
+		}
+		if info.Status == timerStatusPaused {
+			return " " + styleDim + "⏸" + styleReset
+		}
+		remaining := ""
+		if info.FiresAtUnixMS > 0 {
+			d := time.Until(time.UnixMilli(info.FiresAtUnixMS))
+			if d < 0 {
+				d = 0
+			}
+			remaining = formatShortDuration(d)
+		}
+		return " " + styleDim + "⏱" + styleReset + " " + styleHint + remaining + styleReset
+	}
 	statusGlyph := func(c *Child, focused bool) string {
 		if c.Status() != StatusRunning {
 			return styleDim + "○" + styleReset
 		}
+		// Idle-detection states paint over the plain running glyph so
+		// the rail communicates "running but waiting on you" vs "running
+		// and busy" at a glance. Focused entries always use the accent
+		// colour so the user's selection stays visible.
+		style := styleHint
 		if focused {
-			return styleAccent + "●" + styleReset
+			style = styleAccent
+		}
+		switch c.IdleState() {
+		case StateError:
+			return styleError + "✕" + styleReset
+		case StatePermission:
+			return styleAccent + "?" + styleReset
+		case StateThinking:
+			return style + "◐" + styleReset
+		case StateIdle:
+			return style + "○" + styleReset
+		case StateWorking:
+			return style + "●" + styleReset
+		default:
+			return style + "●" + styleReset
 		}
-		return styleHint + "●" + styleReset
 	}
 
 	// Processes section — top-level command/terminal processes,
@@ -88,14 +279,19 @@ func (st *uiState) drawSidebar() {
 		if c.AutoRestart() {
 			marker = " " + styleDim + "⟳" + styleReset
 		}
-		var line string
+		timer := timerIndicator(c)
+		var prefix, openStyle string
 		if focused {
-			line = " " + styleAccent + "▎" + styleReset + " " + glyph + " " +
-				styleBold + c.DisplayName() + styleReset + marker
+			prefix = " " + styleAccent + "▎" + styleReset + " " + glyph + " "
+			openStyle = styleBold
 		} else {
-			line = "   " + glyph + " " + styleHint + c.DisplayName() + styleReset + marker
+			prefix = "   " + glyph + " "
+			openStyle = styleHint
 		}
-		write(line)
+		raw := c.DisplayName()
+		suffix, budget := chooseSidebarSuffix(len([]rune(raw)), width, prefix, marker+timer, timer)
+		nameCell := st.rowNameSlot(c.ID, raw, budget, focused)
+		write(prefix + openStyle + nameCell + styleReset + suffix)
 	}
 
 	// Agent Tree section — formerly "Session tree". Shows the active
@@ -120,67 +316,64 @@ func (st *uiState) drawSidebar() {
 		}
 		focused := c.ID == focus
 		glyph := statusGlyph(c, focused)
-		var line string
+		timer := timerIndicator(c)
+		var prefix, openStyle string
 		if focused {
-			line = " " + styleAccent + "▎" + styleReset + " " + indent + glyph + " " +
-				styleBold + c.DisplayName() + styleReset
+			prefix = " " + styleAccent + "▎" + styleReset + " " + indent + glyph + " "
+			openStyle = styleBold
 		} else {
-			line = "   " + indent + glyph + " " + styleHint + c.DisplayName() + styleReset
+			prefix = "   " + indent + glyph + " "
+			openStyle = styleHint
 		}
-		write(line)
+		raw := c.DisplayName()
+		suffix, budget := chooseSidebarSuffix(len([]rune(raw)), width, prefix, timer, timer)
+		nameCell := st.rowNameSlot(c.ID, raw, budget, focused)
+		write(prefix + openStyle + nameCell + styleReset + suffix)
 	}
 
-	// Scratchpads list — pick the most-recently-modified one as the
-	// preview target. SPEC §4.
-	var previewName string
+	if summary := st.activeSummaryRaw(); summary != "" && row+2 <= maxRow {
+		write("")
+		for _, line := range wrapSidebarSummary(summary, width-4) {
+			if row > maxRow {
+				break
+			}
+			write("   " + styleDim + line + styleReset)
+		}
+	}
+
+	// Scratchpads list — names only. The preview pane used to live
+	// here and clobbered the main viewport when content overflowed the
+	// rail. Focus moves to a pad via Ctrl+W/S; the content renders in
+	// the main viewport via repaintFocusedPad. SPEC §4.
 	if row+2 <= maxRow {
 		write("")
 		writeHeader("Scratchpads")
-		entries, err := st.pads.List()
-		if err == nil {
+		entries := st.padsList()
+		if entries != nil {
 			if len(entries) == 0 {
 				write("   " + styleDim + "(none)" + styleReset)
 			} else {
-				var newestTS string
-				for _, e := range entries {
-					if e.ModifiedAt > newestTS {
-						newestTS = e.ModifiedAt
-						previewName = e.Name
-					}
-				}
 				for _, e := range entries {
 					if row > maxRow {
 						break
 					}
-					var line string
-					if e.Name == previewName {
-						line = " " + styleAccent + "▎" + styleReset + " " +
-							styleBold + e.Name + styleReset
+					focused := e.Name == focusPad
+					var prefix, openStyle string
+					if focused {
+						prefix = " " + styleAccent + "▎" + styleReset + " "
+						openStyle = styleBold
 					} else {
-						line = "   " + styleHint + e.Name + styleReset
+						prefix = "   "
+						openStyle = styleHint
 					}
-					write(line)
+					budget := width - visibleLen(prefix)
+					nameCell := st.rowNameSlot("pad:"+e.Name, e.Name, budget, focused)
+					write(prefix + openStyle + nameCell + styleReset)
 				}
 			}
 		}
 	}
 
-	// Preview pane: dim file content under a thin divider.
-	if previewName != "" && row+2 <= maxRow {
-		write("")
-		write(" " + styleBorder + strings.Repeat("─", width-2) + styleReset)
-		write(" " + styleActive + previewName + styleReset)
-		content, _, err := st.pads.Read(previewName)
-		if err == nil {
-			for _, line := range strings.Split(content, "\n") {
-				if row > maxRow {
-					break
-				}
-				write(" " + styleDim + line + styleReset)
-			}
-		}
-	}
-
 	// Blank-fill any rows the rail content didn't cover so stale
 	// content from a previous redraw doesn't linger.
 	for row <= maxRow {
@@ -191,13 +384,64 @@ func (st *uiState) drawSidebar() {
 	st.chromeCacheMu.Lock()
 	if frame == st.sidebarCache {
 		st.chromeCacheMu.Unlock()
+		if st.metrics != nil {
+			st.metrics.recordSidebar(time.Since(entry), true)
+		}
 		return
 	}
 	st.sidebarCache = frame
 	st.chromeCacheMu.Unlock()
+	if st.metrics != nil {
+		defer func() { st.metrics.recordSidebar(time.Since(entry), false) }()
+	}
 
 	st.outMu.Lock()
 	// Save cursor; emit the sidebar; restore.
 	fmt.Fprintf(os.Stdout, "\x1b7%s\x1b8", frame)
 	st.outMu.Unlock()
 }
+
+func wrapSidebarSummary(s string, width int) []string {
+	if width < 1 {
+		width = 1
+	}
+	words := strings.Fields(s)
+	if len(words) == 0 {
+		return nil
+	}
+	var out []string
+	var cur string
+	for _, word := range words {
+		if visibleLen(word) > width {
+			if cur != "" {
+				out = append(out, cur)
+				cur = ""
+			}
+			for visibleLen(word) > width {
+				out = append(out, clipRunes(word, width))
+				word = string([]rune(word)[width:])
+			}
+			if word != "" {
+				cur = word
+			}
+			continue
+		}
+		if cur == "" {
+			cur = word
+			continue
+		}
+		if visibleLen(cur)+1+visibleLen(word) <= width {
+			cur += " " + word
+			continue
+		}
+		out = append(out, cur)
+		cur = word
+	}
+	if cur != "" {
+		out = append(out, cur)
+	}
+	if len(out) > 3 {
+		out = out[:3]
+	}
+	return out
+}

internal/app/spawn_focus_test.go

@@ -0,0 +1,46 @@
+package app
+
+import (
+	"testing"
+)
+
+// TestOnChildSpawnedAgentChildKeepsFocus verifies that when a child is
+// spawned with a ParentID set (i.e. a patterm-managed agent caused the
+// spawn over MCP), OnChildSpawned does NOT steal viewport focus from
+// the currently focused child.
+func TestOnChildSpawnedAgentChildKeepsFocus(t *testing.T) {
+	sess := NewSession(t.TempDir(), "test")
+	st := &uiState{sess: sess}
+
+	parent := newChildEntry("p_parent", "parent", KindAgent, nil, nil, "", "", "")
+	st.focusedID = parent.ID
+	st.focusedName = parent.Name
+
+	subAgent := newChildEntry("p_sub", "sub", KindAgent, nil, nil, parent.ID, "", "")
+
+	st.OnChildSpawned(subAgent)
+
+	if got := st.focusedID; got != parent.ID {
+		t.Fatalf("agent-initiated spawn should not change focusedID: want %q, got %q", parent.ID, got)
+	}
+	if got := st.focusedName; got != parent.Name {
+		t.Fatalf("focusedName changed: want %q, got %q", parent.Name, got)
+	}
+}
+
+// TestOnChildSpawnedPaletteChildTakesFocus verifies the legacy path is
+// preserved: spawns with an empty ParentID (palette, restore, external
+// MCP caller) still auto-focus the new child.
+func TestOnChildSpawnedPaletteChildTakesFocus(t *testing.T) {
+	sess := NewSession(t.TempDir(), "test")
+	st := &uiState{sess: sess}
+	st.lastExit.Store(-1)
+
+	c := newChildEntry("p_new", "newchild", KindAgent, nil, nil, "", "", "")
+
+	st.OnChildSpawned(c)
+
+	if got := st.focusedID; got != c.ID {
+		t.Fatalf("palette-initiated spawn should auto-focus: want %q, got %q", c.ID, got)
+	}
+}

internal/app/style.go

@@ -11,4 +11,5 @@ const (
 	styleAccent = "\x1b[38;5;75m"
 	styleHint   = "\x1b[38;5;244m"
 	styleActive = "\x1b[1;38;5;253m"
+	styleError  = "\x1b[38;5;203m"
 )

internal/app/summarizer.go

@@ -0,0 +1,463 @@
+package app
+
+import (
+	"bytes"
+	"context"
+	"encoding/json"
+	"fmt"
+	"os/exec"
+	"strings"
+	"sync"
+	"time"
+	"unicode"
+
+	"github.com/hjbdev/patterm/internal/preset"
+)
+
+const (
+	summaryTickInterval = time.Second
+	summaryTimeout      = 90 * time.Second
+	summaryMaxLineCells = 240
+)
+
+type summaryState struct {
+	Text      string
+	State     IdleState
+	UpdatedAt time.Time
+	Error     string
+}
+
+type summaryManager struct {
+	sess       *Session
+	projectDir string
+	presets    preset.Set
+	settings   func() autoSummarySettings
+	onUpdate   func()
+	onResult   func(string, summaryState)
+
+	mu      sync.Mutex
+	tracked map[string]bool
+	entries map[string]*summaryEntry
+}
+
+type summaryEntry struct {
+	armed          bool
+	dirty          bool
+	running        bool
+	lastInputAt    time.Time
+	lastOutputAt   time.Time
+	lastAttemptAt  time.Time
+	lastSummarized int64
+	state          summaryState
+}
+
+type summarizerResponse struct {
+	Summary string `json:"summary"`
+	State   string `json:"state"`
+}
+
+func newSummaryManager(sess *Session, projectDir string, presets preset.Set, settingsFn func() autoSummarySettings, onUpdate func(), onResult func(string, summaryState)) *summaryManager {
+	return &summaryManager{
+		sess:       sess,
+		projectDir: projectDir,
+		presets:    presets,
+		settings:   settingsFn,
+		onUpdate:   onUpdate,
+		onResult:   onResult,
+		tracked:    make(map[string]bool),
+		entries:    make(map[string]*summaryEntry),
+	}
+}
+
+func (m *summaryManager) run(ctx context.Context) {
+	ticker := time.NewTicker(summaryTickInterval)
+	defer ticker.Stop()
+	for {
+		select {
+		case <-ctx.Done():
+			return
+		case <-ticker.C:
+			m.maybeStart(ctx, time.Now())
+		}
+	}
+}
+
+func (m *summaryManager) ObserveHumanInput(childID string, b []byte) {
+	if m == nil || !m.isTracked(childID) {
+		return
+	}
+	cfg := m.settings()
+	if len(strings.TrimSpace(string(b))) < cfg.MinInputChars {
+		return
+	}
+	m.mu.Lock()
+	e := m.entryLocked(childID)
+	e.armed = true
+	e.lastInputAt = time.Now()
+	m.mu.Unlock()
+}
+
+func (m *summaryManager) ObserveOutput(childID string) {
+	if m == nil || !m.isTracked(childID) {
+		return
+	}
+	m.mu.Lock()
+	e := m.entryLocked(childID)
+	if e.armed {
+		e.dirty = true
+		e.lastOutputAt = time.Now()
+	}
+	m.mu.Unlock()
+}
+
+func (m *summaryManager) RegisterChild(c *Child) {
+	if m == nil || c == nil {
+		return
+	}
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	if isTopLevelSummarizedAgent(c) {
+		m.tracked[c.ID] = true
+	} else {
+		delete(m.tracked, c.ID)
+	}
+}
+
+func (m *summaryManager) UnregisterChild(id string) {
+	if m == nil || id == "" {
+		return
+	}
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	delete(m.tracked, id)
+}
+
+func (m *summaryManager) isTracked(id string) bool {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	return m.tracked[id]
+}
+
+func (m *summaryManager) Summary(childID string) summaryState {
+	if m == nil || childID == "" {
+		return summaryState{}
+	}
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	if e := m.entries[childID]; e != nil {
+		return e.state
+	}
+	return summaryState{}
+}
+
+func (m *summaryManager) RunNow(ctx context.Context, childID string) {
+	if m == nil || childID == "" {
+		return
+	}
+	c := m.sess.FindChild(childID)
+	if !isTopLevelSummarizedAgent(c) {
+		return
+	}
+	m.mu.Lock()
+	e := m.entryLocked(c.ID)
+	if e.running {
+		m.mu.Unlock()
+		return
+	}
+	e.running = true
+	e.lastAttemptAt = time.Now()
+	m.mu.Unlock()
+	go m.runOne(ctx, c.ID, true)
+}
+
+func (m *summaryManager) Test(ctx context.Context) error {
+	cfg := m.settings()
+	return runSummarizerHealth(ctx, cfg, m.projectDir)
+}
+
+func (m *summaryManager) entryLocked(id string) *summaryEntry {
+	e := m.entries[id]
+	if e == nil {
+		e = &summaryEntry{}
+		m.entries[id] = e
+	}
+	return e
+}
+
+func (m *summaryManager) maybeStart(ctx context.Context, now time.Time) {
+	cfg := m.settings()
+	if !cfg.Enabled {
+		return
+	}
+	cadence, err := time.ParseDuration(cfg.Cadence)
+	if err != nil || cadence <= 0 {
+		cadence = time.Minute
+	}
+	quiet := time.Duration(cfg.QuietWindowMS) * time.Millisecond
+	var startID string
+	for _, c := range m.sess.Children() {
+		if !isTopLevelSummarizedAgent(c) {
+			continue
+		}
+		m.mu.Lock()
+		e := m.entryLocked(c.ID)
+		eligible := e.armed && e.dirty && !e.running &&
+			!e.lastOutputAt.IsZero() && now.Sub(e.lastOutputAt) >= quiet &&
+			(e.lastAttemptAt.IsZero() || now.Sub(e.lastAttemptAt) >= cadence) &&
+			c.ScreenVersion() != e.lastSummarized
+		if eligible {
+			e.running = true
+			e.lastAttemptAt = now
+			startID = c.ID
+		}
+		m.mu.Unlock()
+		if startID != "" {
+			go m.runOne(ctx, startID, false)
+			return
+		}
+	}
+}
+
+func (m *summaryManager) runOne(ctx context.Context, childID string, manual bool) {
+	c := m.sess.FindChild(childID)
+	if c == nil {
+		m.finish(childID, summaryState{Error: "process disappeared"}, 0)
+		return
+	}
+	cfg := m.settings()
+	snapshot := buildSummarySnapshot(c, cfg.MaxHistoryChars, m.chromeHintsFor(c.PresetRef))
+	if strings.TrimSpace(snapshot) == "" {
+		m.finish(childID, summaryState{Error: "empty snapshot"}, c.ScreenVersion())
+		return
+	}
+	runCtx, cancel := context.WithTimeout(ctx, summaryTimeout)
+	defer cancel()
+	resp, err := runSummarizer(runCtx, cfg, m.projectDir, snapshot)
+	st := summaryState{UpdatedAt: time.Now()}
+	if err != nil {
+		st.Error = err.Error()
+		m.finish(childID, st, c.ScreenVersion())
+		return
+	}
+	st.Text = strings.TrimSpace(resp.Summary)
+	st.State = summaryIdleState(resp.State)
+	if st.Text == "" {
+		st.Error = "empty summary"
+	}
+	if manual && st.Text != "" && st.State == StateUnknown {
+		st.State = c.IdleState()
+	}
+	m.finish(childID, st, c.ScreenVersion())
+}
+
+func (m *summaryManager) finish(childID string, st summaryState, version int64) {
+	m.mu.Lock()
+	e := m.entryLocked(childID)
+	e.running = false
+	if st.Text != "" || st.Error != "" {
+		if st.Text == "" && e.state.Text != "" {
+			st.Text = e.state.Text
+			st.State = e.state.State
+			st.UpdatedAt = e.state.UpdatedAt
+		}
+		e.state = st
+	}
+	if st.Text != "" {
+		e.armed = false
+		e.dirty = false
+		e.lastSummarized = version
+	}
+	m.mu.Unlock()
+	if m.onUpdate != nil {
+		m.onUpdate()
+	}
+	if m.onResult != nil && (st.Text != "" || st.Error != "") {
+		m.onResult(childID, st)
+	}
+}
+
+func isTopLevelSummarizedAgent(c *Child) bool {
+	return c != nil && c.Kind == KindAgent && c.ParentID == "" && c.Status() == StatusRunning
+}
+
+func (m *summaryManager) chromeHintsFor(presetName string) []string {
+	if presetName == "" {
+		return nil
+	}
+	for _, p := range m.presets.Agents {
+		if p.Name == presetName {
+			return p.ChromeTrimHints
+		}
+	}
+	return nil
+}
+
+func buildSummarySnapshot(c *Child, maxChars int, chromeHints []string) string {
+	if maxChars <= 0 {
+		maxChars = 12000
+	}
+	grid := ""
+	if em := c.Emulator(); em != nil {
+		if txt, err := em.PlainText(); err == nil {
+			grid = compactSummaryText(applyChromeTrim(txt, chromeHints))
+		}
+	}
+	tailBytes := max(maxChars*4, maxChars)
+	b := c.tailBytes(tailBytes)
+	history := compactSummaryText(applyChromeTrim(string(stripANSIBytes(nil, b)), chromeHints))
+	history = tailString(history, maxChars)
+	var out strings.Builder
+	if history != "" {
+		out.WriteString("Recent rendered history:\n")
+		out.WriteString(history)
+		out.WriteString("\n\n")
+	}
+	if grid != "" && !strings.Contains(history, grid) {
+		out.WriteString("Current visible grid:\n")
+		out.WriteString(grid)
+	}
+	return tailString(out.String(), maxChars)
+}
+
+func compactSummaryText(in string) string {
+	in = string(stripANSIBytes(nil, []byte(in)))
+	in = strings.ReplaceAll(in, "\r\n", "\n")
+	in = strings.ReplaceAll(in, "\r", "\n")
+	lines := strings.Split(in, "\n")
+	out := make([]string, 0, len(lines))
+	blank := false
+	for _, line := range lines {
+		line = strings.TrimRightFunc(line, unicode.IsSpace)
+		line = strings.Map(func(r rune) rune {
+			if r == '\t' || r == '\n' {
+				return r
+			}
+			if r < 0x20 || r == 0x7f {
+				return -1
+			}
+			return r
+		}, line)
+		line = truncateSummaryLine(line, summaryMaxLineCells)
+		if strings.TrimSpace(line) == "" {
+			if blank {
+				continue
+			}
+			blank = true
+			out = append(out, "")
+			continue
+		}
+		blank = false
+		out = append(out, line)
+	}
+	return strings.TrimSpace(strings.Join(out, "\n"))
+}
+
+func truncateSummaryLine(s string, max int) string {
+	if max <= 0 || visibleLen(s) <= max {
+		return s
+	}
+	return clipRunes(s, max-1) + "…"
+}
+
+func tailString(s string, max int) string {
+	rs := []rune(s)
+	if len(rs) <= max {
+		return s
+	}
+	return string(rs[len(rs)-max:])
+}
+
+func runSummarizer(ctx context.Context, cfg autoSummarySettings, projectDir, snapshot string) (summarizerResponse, error) {
+	prompt := summaryPrompt(snapshot)
+	out, err := runSummarizerCommand(ctx, cfg, projectDir, prompt)
+	if err != nil {
+		return summarizerResponse{}, err
+	}
+	resp, err := parseSummarizerResponse(out)
+	if err != nil {
+		return summarizerResponse{}, err
+	}
+	if summaryIdleState(resp.State) == StateUnknown {
+		return summarizerResponse{}, fmt.Errorf("invalid summary state %q", resp.State)
+	}
+	return resp, nil
+}
+
+func runSummarizerHealth(ctx context.Context, cfg autoSummarySettings, projectDir string) error {
+	out, err := runSummarizerCommand(ctx, cfg, projectDir, "Reply with exactly: patterm okay")
+	if err != nil {
+		return err
+	}
+	if strings.TrimSpace(out) != "patterm okay" {
+		return fmt.Errorf("health check did not return patterm okay")
+	}
+	return nil
+}
+
+func runSummarizerCommand(ctx context.Context, cfg autoSummarySettings, projectDir, prompt string) (string, error) {
+	provider := cfg.Provider
+	model := cfg.modelFor(provider)
+	var cmd *exec.Cmd
+	switch provider {
+	case "opencode":
+		cmd = exec.CommandContext(ctx, "opencode", "run", "--model", model, "--dir", projectDir, prompt)
+	case "claude":
+		cmd = exec.CommandContext(ctx, "claude", "--print", "--model", model, prompt)
+	default:
+		cmd = exec.CommandContext(ctx, "codex", "exec", "--ephemeral", "--skip-git-repo-check", "--sandbox", "read-only", "--model", model, "-")
+		cmd.Stdin = strings.NewReader(prompt)
+	}
+	cmd.Dir = projectDir
+	var stderr bytes.Buffer
+	cmd.Stderr = &stderr
+	out, err := cmd.Output()
+	if err != nil {
+		msg := strings.TrimSpace(stderr.String())
+		if msg == "" {
+			msg = err.Error()
+		}
+		return "", fmt.Errorf("%s summarizer: %s", provider, msg)
+	}
+	return string(out), nil
+}
+
+func summaryPrompt(snapshot string) string {
+	return "Summarize this terminal/agent snapshot for a compact UI catch-up aid.\n" +
+		"Return only JSON with keys summary and state. State must be one of IDLE, PERMISSION, THINKING, WORKING, ERROR.\n" +
+		"Keep summary under 180 characters, concrete, and avoid mentioning that you are summarizing.\n\n" +
+		snapshot
+}
+
+func parseSummarizerResponse(out string) (summarizerResponse, error) {
+	var resp summarizerResponse
+	if err := json.Unmarshal([]byte(strings.TrimSpace(out)), &resp); err == nil {
+		return resp, nil
+	}
+	for _, line := range strings.Split(out, "\n") {
+		line = strings.TrimSpace(line)
+		if !strings.HasPrefix(line, "{") || !strings.HasSuffix(line, "}") {
+			continue
+		}
+		if err := json.Unmarshal([]byte(line), &resp); err == nil {
+			return resp, nil
+		}
+	}
+	return resp, fmt.Errorf("summary output was not JSON")
+}
+
+func summaryIdleState(s string) IdleState {
+	switch strings.ToUpper(strings.TrimSpace(s)) {
+	case "IDLE":
+		return StateIdle
+	case "PERMISSION":
+		return StatePermission
+	case "THINKING":
+		return StateThinking
+	case "WORKING":
+		return StateWorking
+	case "ERROR":
+		return StateError
+	default:
+		return StateUnknown
+	}
+}

internal/app/summarizer_test.go

@@ -0,0 +1,125 @@
+package app
+
+import (
+	"strings"
+	"testing"
+
+	"github.com/hjbdev/patterm/internal/preset"
+)
+
+func TestParseSummarizerResponseAllowsWrappedJSON(t *testing.T) {
+	resp, err := parseSummarizerResponse("log\n{\"summary\":\"Waiting for tests\",\"state\":\"WORKING\"}\n")
+	if err != nil {
+		t.Fatalf("parseSummarizerResponse: %v", err)
+	}
+	if resp.Summary != "Waiting for tests" || summaryIdleState(resp.State) != StateWorking {
+		t.Fatalf("response = %+v", resp)
+	}
+}
+
+func TestCompactSummaryTextDropsControlAndRedundantWhitespace(t *testing.T) {
+	got := compactSummaryText("hello\x00 world  \n\n\n\x1b[31mred\x1b[0m\n")
+	if strings.ContainsRune(got, '\x00') {
+		t.Fatalf("control byte survived: %q", got)
+	}
+	if strings.Contains(got, "\n\n\n") {
+		t.Fatalf("redundant blanks survived: %q", got)
+	}
+	if strings.Contains(got, "\x1b") {
+		t.Fatalf("ansi survived: %q", got)
+	}
+}
+
+func TestWrapSidebarSummaryKeepsWordBoundaries(t *testing.T) {
+	got := wrapSidebarSummary("alpha beta gamma delta", 12)
+	want := []string{"alpha beta", "gamma delta"}
+	if len(got) != len(want) {
+		t.Fatalf("lines = %#v", got)
+	}
+	for i := range want {
+		if got[i] != want[i] {
+			t.Fatalf("line %d = %q want %q", i, got[i], want[i])
+		}
+	}
+	long := wrapSidebarSummary("supercalifragilistic short", 8)
+	if len(long) == 0 || strings.Contains(strings.Join(long, ""), "…") {
+		t.Fatalf("long word should wrap without ellipsis: %#v", long)
+	}
+	for _, line := range long {
+		if visibleLen(line) > 8 {
+			t.Fatalf("line %q exceeds width", line)
+		}
+	}
+}
+
+func TestSummaryTextForSelectsChildAndClips(t *testing.T) {
+	sess := NewSession(t.TempDir(), "test")
+	cfg := defaultSettings()
+	st := &uiState{
+		sess:     sess,
+		settings: cfg,
+		summaries: newSummaryManager(sess, t.TempDir(), preset.Set{}, func() autoSummarySettings {
+			return cfg.AutoSummary.clone()
+		}, nil, nil),
+	}
+	st.summaries.mu.Lock()
+	st.summaries.entries["a1"] = &summaryEntry{state: summaryState{Text: " alpha summary "}}
+	st.summaries.entries["a2"] = &summaryEntry{state: summaryState{Text: "beta summary"}}
+	st.summaries.entries["empty"] = &summaryEntry{state: summaryState{Text: "   "}}
+	st.summaries.entries["long"] = &summaryEntry{state: summaryState{Text: "abcdefghijklmnopqrstuvwxyz"}}
+	st.summaries.mu.Unlock()
+
+	if got := st.summaryTextFor("a2", 20); got != "beta summary" {
+		t.Fatalf("summaryTextFor(a2) = %q, want beta summary", got)
+	}
+	if got := st.summaryTextFor("empty", 20); got != "" {
+		t.Fatalf("summaryTextFor(empty) = %q, want empty", got)
+	}
+	if got := st.summaryTextFor("long", 8); got != "abcdefg…" {
+		t.Fatalf("summaryTextFor(long) = %q, want abcdefg…", got)
+	}
+
+	st.settingsMu.Lock()
+	st.settings.AutoSummary.Enabled = false
+	st.settingsMu.Unlock()
+	if got := st.summaryTextFor("a1", 20); got != "" {
+		t.Fatalf("summaryTextFor disabled = %q, want empty", got)
+	}
+}
+
+func TestSummaryManagerArmsOnlyTrackedTopLevelAgents(t *testing.T) {
+	sess := NewSession(t.TempDir(), "test")
+	c := newChildEntry("a1", "agent", KindAgent, []string{"fake"}, nil, "", "", "")
+	running := StatusRunning
+	c.status.Store(&running)
+	sess.children[c.ID] = c
+	sess.order = append(sess.order, c.ID)
+	cfg := defaultSettings().AutoSummary
+	m := newSummaryManager(sess, t.TempDir(), preset.Set{}, func() autoSummarySettings {
+		return cfg.clone()
+	}, nil, nil)
+	m.ObserveHumanInput(c.ID, []byte("please summarize"))
+	if got := m.Summary(c.ID); got.Text != "" {
+		t.Fatalf("untracked agent should not update summary state: %+v", got)
+	}
+	m.RegisterChild(c)
+	m.ObserveHumanInput(c.ID, []byte("please summarize"))
+	m.ObserveOutput(c.ID)
+	m.mu.Lock()
+	e := m.entries[c.ID]
+	m.mu.Unlock()
+	if e == nil || !e.armed || !e.dirty {
+		t.Fatalf("tracked top-level agent not armed/dirty: %+v", e)
+	}
+
+	sub := newChildEntry("a2", "sub", KindAgent, []string{"fake"}, nil, c.ID, "", "")
+	sub.status.Store(&running)
+	m.RegisterChild(sub)
+	m.ObserveHumanInput(sub.ID, []byte("please summarize"))
+	m.mu.Lock()
+	_, ok := m.entries[sub.ID]
+	m.mu.Unlock()
+	if ok {
+		t.Fatal("sub-agent should not get a summary entry")
+	}
+}

internal/app/tabbar.go

@@ -4,12 +4,13 @@ import (
 	"fmt"
 	"os"
 	"strings"
+	"time"
 	"unicode/utf8"
 )
 
-// Two-row tab bar: labels row, underline row. The PTY viewport's top
+// Three-row tab bar: labels row, active-thread summary row, underline row. The PTY viewport's top
 // row is therefore mainTop == tabBarRows + 1.
-const tabBarRows = 2
+const tabBarRows = 3
 
 // drawTabBar renders the top tab strip across the full host width.
 // Tabs share the available width with a flex layout — each visible
@@ -17,9 +18,17 @@ const tabBarRows = 2
 // to the leftmost tabs so the strip fills the screen edge-to-edge.
 // A trailing "+ new" hint sits in the rightmost reserved slot.
 func (st *uiState) drawTabBar() {
+	var entry time.Time
+	if st.metrics != nil {
+		entry = time.Now()
+	}
 	st.mu.Lock()
 	palOpen := st.palette != nil
-	focus := st.focusedID
+	// Highlight the top-level agent tab even when focus has stepped
+	// into a sub-agent (or a Processes pane entry). activeAgentID walks
+	// the parent chain to the root, so the user always sees which tab
+	// their current thread belongs to.
+	focus := st.activeAgentID
 	st.mu.Unlock()
 	if palOpen {
 		return
@@ -50,12 +59,14 @@ func (st *uiState) drawTabBar() {
 	newHintW := utf8.RuneCountInString(newHint) + 2 // " + new " framing
 
 	type tabRect struct {
-		startCol int
-		width    int
-		label    string
-		active   bool
+		childID    string
+		startCol   int
+		width      int
+		label      string
+		glyph      string
+		glyphStyle string
+		active     bool
 	}
-
 	// Reserve space at the right edge for "+ new". If there are too
 	// many tabs to fit even at minTabWidth, drop tabs from the right
 	// until they do. The current focus stays visible.
@@ -105,9 +116,16 @@ func (st *uiState) drawTabBar() {
 			if i < extra {
 				w++
 			}
+			active := c.ID == focus
+			glyph, glyphStyle := tabIdleGlyph(c.IdleState(), active)
 			label := c.DisplayName()
 			labelW := utf8.RuneCountInString(label)
-			maxLabelW := w - 2 // one pad on each side
+			// Reserve room for the glyph + its trailing space when present
+			// (1 + 1 runes), on top of the one-cell pad on each side.
+			maxLabelW := w - 2
+			if glyph != "" {
+				maxLabelW -= 2
+			}
 			if maxLabelW < 1 {
 				maxLabelW = 1
 			}
@@ -120,17 +138,21 @@ func (st *uiState) drawTabBar() {
 				labelW = utf8.RuneCountInString(label)
 			}
 			tabs = append(tabs, tabRect{
-				startCol: col,
-				width:    w,
-				label:    label,
-				active:   c.ID == focus,
+				childID:    c.ID,
+				startCol:   col,
+				width:      w,
+				label:      label,
+				glyph:      glyph,
+				glyphStyle: glyphStyle,
+				active:     active,
 			})
 			col += w
 		}
 	}
 
 	var b strings.Builder
-	// Clear both rows so a stale label from the previous frame can't
+	// Clear all tab-bar rows so stale labels or summaries from the
+	// previous frame can't
 	// bleed through. Use ECH clamped to `width` (= childCols) instead of
 	// `\x1b[2K`: 2K wipes the entire line including the sidebar columns,
 	// and if drawSidebar's chrome cache is fresh it won't repaint to
@@ -138,32 +160,47 @@ func (st *uiState) drawTabBar() {
 	// and content should be.
 	fmt.Fprintf(&b, "\x1b[1;1H\x1b[%dX", width)
 	fmt.Fprintf(&b, "\x1b[2;1H\x1b[%dX", width)
+	fmt.Fprintf(&b, "\x1b[3;1H\x1b[%dX", width)
 
 	for _, t := range tabs {
-		// Row 1: centre-ish label inside the tab cell.
+		// Row 1: centre-ish glyph+label inside the tab cell.
 		labelW := utf8.RuneCountInString(t.label)
-		leftPad := (t.width - labelW) / 2
+		visibleW := labelW
+		if t.glyph != "" {
+			visibleW += 2 // glyph + separator space
+		}
+		leftPad := (t.width - visibleW) / 2
 		if leftPad < 1 {
 			leftPad = 1
 		}
-		rightPad := t.width - labelW - leftPad
+		rightPad := t.width - visibleW - leftPad
 		if rightPad < 0 {
 			rightPad = 0
 		}
-		fmt.Fprintf(&b, "\x1b[1;%dH", t.startCol)
+		cellStyle := styleHint
 		if t.active {
-			b.WriteString(styleActive)
-		} else {
-			b.WriteString(styleHint)
+			cellStyle = styleActive
 		}
+		fmt.Fprintf(&b, "\x1b[1;%dH", t.startCol)
+		b.WriteString(cellStyle)
 		b.WriteString(strings.Repeat(" ", leftPad))
+		if t.glyph != "" {
+			// Glyph uses its own colour so error/permission states pop
+			// regardless of tab focus, matching the sidebar's vocabulary.
+			b.WriteString(styleReset)
+			b.WriteString(t.glyphStyle)
+			b.WriteString(t.glyph)
+			b.WriteString(styleReset)
+			b.WriteString(cellStyle)
+			b.WriteString(" ")
+		}
 		b.WriteString(t.label)
 		b.WriteString(strings.Repeat(" ", rightPad))
 		b.WriteString(styleReset)
 
-		// Row 2: underline. Thick accent for the active tab, faint
+		// Row 3: underline. Thick accent for the active tab, faint
 		// border for the rest.
-		fmt.Fprintf(&b, "\x1b[2;%dH", t.startCol)
+		fmt.Fprintf(&b, "\x1b[3;%dH", t.startCol)
 		if t.active {
 			b.WriteString(styleAccent)
 			b.WriteString(strings.Repeat("━", t.width))
@@ -180,20 +217,59 @@ func (st *uiState) drawTabBar() {
 		fmt.Fprintf(&b, "\x1b[1;%dH %s%s%s ", hintCol, styleDim, newHint, styleReset)
 		// Underline continues faintly under the hint so the strip
 		// reads as one bar.
-		fmt.Fprintf(&b, "\x1b[2;%dH%s%s%s",
+		fmt.Fprintf(&b, "\x1b[3;%dH%s%s%s",
 			hintCol, styleBorder, strings.Repeat("─", newHintW), styleReset)
 	}
 
+	for _, tab := range tabs {
+		summaryWidth := tab.width - 2
+		if summary := st.summaryTextFor(tab.childID, summaryWidth); summary != "" {
+			fmt.Fprintf(&b, "\x1b[2;%dH %s%s%s", tab.startCol, styleDim, summary, styleReset)
+		}
+	}
+
 	frame := b.String()
 	st.chromeCacheMu.Lock()
 	if frame == st.tabBarCache {
 		st.chromeCacheMu.Unlock()
+		if st.metrics != nil {
+			st.metrics.recordTabbar(time.Since(entry), true)
+		}
 		return
 	}
 	st.tabBarCache = frame
 	st.chromeCacheMu.Unlock()
+	if st.metrics != nil {
+		defer func() { st.metrics.recordTabbar(time.Since(entry), false) }()
+	}
 
 	st.outMu.Lock()
 	defer st.outMu.Unlock()
 	fmt.Fprintf(os.Stdout, "\x1b7%s\x1b8", frame)
 }
+
+// tabIdleGlyph returns the one-rune state indicator (and its SGR style)
+// to render before a tab's label. Mirrors the sidebar's vocabulary so
+// users learn the symbols in one place: ✕ error, ? permission, ◐
+// thinking, ○ idle, ● working. Returns ("", "") for StateUnknown so the
+// first frame after spawn doesn't show a misleading badge.
+func tabIdleGlyph(state IdleState, active bool) (string, string) {
+	base := styleHint
+	if active {
+		base = styleAccent
+	}
+	switch state {
+	case StateError:
+		return "✕", styleError
+	case StatePermission:
+		return "?", styleAccent
+	case StateThinking:
+		return "◐", base
+	case StateIdle:
+		return "○", base
+	case StateWorking:
+		return "●", base
+	default:
+		return "", ""
+	}
+}

internal/app/timers.go

@@ -0,0 +1,686 @@
+package app
+
+import (
+	"fmt"
+	"sync"
+	"time"
+
+	"github.com/hjbdev/patterm/internal/mcp"
+)
+
+// pendingTimerKind picks the firing rule.
+type pendingTimerKind string
+
+const (
+	timerKindDelay   pendingTimerKind = "delay"
+	timerKindIdleAny pendingTimerKind = "idle_any"
+	timerKindIdleAll pendingTimerKind = "idle_all"
+)
+
+const (
+	timerStatusPending  = "pending"
+	timerStatusPaused   = "paused"
+	timerStatusFired    = "fired"
+	timerStatusCanceled = "canceled"
+)
+
+// pendingTimer is one live timer tracked by the manager. The body is
+// delivered verbatim to the owning child's PTY as a fresh user turn
+// when the timer fires.
+//
+// Locking: every field is protected by timerManager.mu. The runtime
+// time.Timer (rt) is started outside the lock so the firing goroutine
+// can take the lock without deadlocking.
+type pendingTimer struct {
+	id      string
+	label   string
+	body    string
+	ownerID string
+	kind    pendingTimerKind
+	status  string
+
+	watched      []string
+	idleBaseline map[string]bool // for idle_any: ids already idle at registration (excluded from satisfaction)
+
+	firesAt          time.Time
+	pausedRemaining  time.Duration
+	pausedWasMaxWait bool // for idle_*: true if the active timer was max-wait, not delay
+
+	rt *time.Timer // delay timer or idle_* max-wait fallback
+}
+
+// timerManager owns the pending-timer registry. Mutating operations
+// (set, cancel, pause, resume) all serialise through mu; fire callbacks
+// from the runtime timer also take mu to safely transition state.
+type timerManager struct {
+	sess *Session
+
+	mu      sync.Mutex
+	nextID  int
+	timers  map[string]*pendingTimer
+	changes chan struct{}
+
+	// fireFn is the callback used to deliver the body to the owning
+	// process. Decoupled so tests can substitute a recorder. Defaults
+	// to caller.InjectAsOrchestrator + "\r".
+	fireFn func(owner *Child, body, label string)
+}
+
+func newTimerManager(sess *Session) *timerManager {
+	m := &timerManager{
+		sess:    sess,
+		timers:  make(map[string]*pendingTimer),
+		changes: make(chan struct{}, 1),
+	}
+	m.fireFn = defaultFireFn
+	return m
+}
+
+func (m *timerManager) changeEvents() <-chan struct{} {
+	return m.changes
+}
+
+func (m *timerManager) notifyChanged() {
+	select {
+	case m.changes <- struct{}{}:
+	default:
+	}
+}
+
+func defaultFireFn(owner *Child, body, label string) {
+	if owner == nil || !owner.IsLive() {
+		return
+	}
+	// Solo delivers body verbatim. patterm's PTY-injection path expects
+	// a trailing CR so the line submits in TUI agents (Claude/Codex/
+	// OpenCode all paste-detect). A bare body without CR sits in the
+	// input buffer; that's almost never what the caller wants.
+	if body == "" {
+		body = fmt.Sprintf("[system] Your timer [%s] has completed.", label)
+	}
+	_ = owner.InjectAsOrchestrator([]byte(body + "\r"))
+}
+
+func (m *timerManager) mintID() string {
+	m.nextID++
+	return fmt.Sprintf("t%d", m.nextID)
+}
+
+// TimerSet schedules a delay timer. Returns immediately; the body is
+// delivered to the owning child when the timer fires.
+func (m *timerManager) TimerSet(ownerID string, body, label string, seconds float64) (string, error) {
+	owner := m.sess.FindChild(ownerID)
+	if owner == nil {
+		return "", mcp.Errorf(mcp.ErrorKindNotFound, "caller %q not known to patterm", ownerID)
+	}
+	if seconds < 0 {
+		return "", mcp.Errorf(mcp.ErrorKindInvalidArgs, "timer_set: seconds must be ≥ 0")
+	}
+	d := time.Duration(seconds * float64(time.Second))
+	m.mu.Lock()
+	id := m.mintID()
+	if label == "" {
+		label = id
+	}
+	t := &pendingTimer{
+		id:      id,
+		label:   label,
+		body:    body,
+		ownerID: ownerID,
+		kind:    timerKindDelay,
+		status:  timerStatusPending,
+		firesAt: time.Now().Add(d),
+	}
+	m.timers[id] = t
+	m.mu.Unlock()
+	t.rt = time.AfterFunc(d, func() { m.fireDelay(id) })
+	m.notifyChanged()
+	return id, nil
+}
+
+func (m *timerManager) fireDelay(id string) {
+	m.mu.Lock()
+	t, ok := m.timers[id]
+	if !ok || t.status != timerStatusPending {
+		m.mu.Unlock()
+		return
+	}
+	t.status = timerStatusFired
+	owner := m.sess.FindChild(t.ownerID)
+	body, label := t.body, t.label
+	delete(m.timers, id)
+	m.mu.Unlock()
+	m.notifyChanged()
+	m.fireFn(owner, body, label)
+}
+
+// TimerFireWhenIdleAny schedules an idle-any timer. Children already
+// idle at registration are excluded from satisfaction — only a
+// transition into idle by a still-active watched child fires the
+// timer. Max-wait, when positive, acts as a fallback fire deadline.
+func (m *timerManager) TimerFireWhenIdleAny(ownerID, body, label string, watched []string, maxWait float64) (mcp.TimerFireWhenIdleResponse, error) {
+	return m.registerIdleTimer(timerKindIdleAny, ownerID, body, label, watched, maxWait)
+}
+
+// TimerFireWhenIdleAll schedules an idle-all timer. Already-idle
+// children count as satisfied; if every watched child is already idle
+// at registration time the response is "already_satisfied" with no
+// timer created.
+func (m *timerManager) TimerFireWhenIdleAll(ownerID, body, label string, watched []string, maxWait float64) (mcp.TimerFireWhenIdleResponse, error) {
+	return m.registerIdleTimer(timerKindIdleAll, ownerID, body, label, watched, maxWait)
+}
+
+func (m *timerManager) registerIdleTimer(kind pendingTimerKind, ownerID, body, label string, watched []string, maxWait float64) (mcp.TimerFireWhenIdleResponse, error) {
+	if m.sess.FindChild(ownerID) == nil {
+		return mcp.TimerFireWhenIdleResponse{}, mcp.Errorf(mcp.ErrorKindNotFound, "caller %q not known to patterm", ownerID)
+	}
+	if len(watched) == 0 {
+		return mcp.TimerFireWhenIdleResponse{}, mcp.Errorf(mcp.ErrorKindInvalidArgs, "watched must contain at least one process_id")
+	}
+	if maxWait < 0 {
+		return mcp.TimerFireWhenIdleResponse{}, mcp.Errorf(mcp.ErrorKindInvalidArgs, "max_wait_seconds must be ≥ 0")
+	}
+	// Validate watched ids and compute the idle baseline up front.
+	already := make([]string, 0)
+	waiting := make([]string, 0)
+	baseline := make(map[string]bool, len(watched))
+	for _, id := range watched {
+		c := m.sess.FindChild(id)
+		if c == nil {
+			return mcp.TimerFireWhenIdleResponse{}, mcp.Errorf(mcp.ErrorKindNotFound, "no such process %q in watched", id)
+		}
+		if isIdleState(c.IdleState()) {
+			already = append(already, id)
+			baseline[id] = true
+		} else {
+			waiting = append(waiting, id)
+		}
+	}
+	resp := mcp.TimerFireWhenIdleResponse{AlreadyIdle: already, WaitingOn: waiting}
+
+	// idle_all: if all watched are already idle, satisfy synchronously
+	// — Solo semantics; no pending timer is created.
+	if kind == timerKindIdleAll && len(waiting) == 0 {
+		resp.Status = "already_satisfied"
+		owner := m.sess.FindChild(ownerID)
+		go m.fireFn(owner, body, label)
+		return resp, nil
+	}
+
+	m.mu.Lock()
+	id := m.mintID()
+	if label == "" {
+		label = id
+	}
+	t := &pendingTimer{
+		id:           id,
+		label:        label,
+		body:         body,
+		ownerID:      ownerID,
+		kind:         kind,
+		status:       timerStatusPending,
+		watched:      append([]string(nil), watched...),
+		idleBaseline: baseline,
+	}
+	if maxWait > 0 {
+		d := time.Duration(maxWait * float64(time.Second))
+		t.firesAt = time.Now().Add(d)
+		t.rt = time.AfterFunc(d, func() { m.fireIdleMaxWait(id) })
+	}
+	m.timers[id] = t
+	m.mu.Unlock()
+	m.notifyChanged()
+	resp.ID = id
+	resp.Status = "pending"
+	return resp, nil
+}
+
+func (m *timerManager) fireIdleMaxWait(id string) {
+	m.mu.Lock()
+	t, ok := m.timers[id]
+	if !ok || t.status != timerStatusPending {
+		m.mu.Unlock()
+		return
+	}
+	t.status = timerStatusFired
+	owner := m.sess.FindChild(t.ownerID)
+	body, label := t.body, t.label
+	delete(m.timers, id)
+	m.mu.Unlock()
+	m.notifyChanged()
+	m.fireFn(owner, body, label)
+}
+
+// onChildStateChanged evaluates every pending idle_any / idle_all
+// timer whenever any child's IdleState flips. Cheap — there are few
+// pending timers and the per-tick check is just a map lookup + a slice
+// scan.
+func (m *timerManager) onChildStateChanged(childID string, state IdleState) {
+	if !isIdleState(state) {
+		return
+	}
+	m.mu.Lock()
+	type firing struct {
+		owner *Child
+		body  string
+		label string
+	}
+	var fires []firing
+	var firedIDs []string
+	for _, t := range m.timers {
+		if t.status != timerStatusPending {
+			continue
+		}
+		if !contains(t.watched, childID) {
+			continue
+		}
+		switch t.kind {
+		case timerKindIdleAny:
+			if t.idleBaseline[childID] {
+				continue // already idle at registration; excluded
+			}
+			t.status = timerStatusFired
+			if t.rt != nil {
+				t.rt.Stop()
+			}
+			fires = append(fires, firing{
+				owner: m.sess.FindChild(t.ownerID),
+				body:  t.body,
+				label: t.label,
+			})
+			firedIDs = append(firedIDs, t.id)
+		case timerKindIdleAll:
+			if m.allWatchedIdleLocked(t) {
+				t.status = timerStatusFired
+				if t.rt != nil {
+					t.rt.Stop()
+				}
+				fires = append(fires, firing{
+					owner: m.sess.FindChild(t.ownerID),
+					body:  t.body,
+					label: t.label,
+				})
+				firedIDs = append(firedIDs, t.id)
+			}
+		}
+	}
+	for _, id := range firedIDs {
+		delete(m.timers, id)
+	}
+	m.mu.Unlock()
+	if len(firedIDs) > 0 {
+		m.notifyChanged()
+	}
+	for _, f := range fires {
+		m.fireFn(f.owner, f.body, f.label)
+	}
+}
+
+// onChildClosed drops pending timer references to childID. Called
+// from Session.Close (and the terminal-corpse cleanup in reapChild)
+// via the session listener bus — a deliberate signal from the host
+// that childID is gone and the parent is not waiting on it anymore.
+//
+// Semantics:
+//   - timers owned by childID are cancelled and deleted: their owner
+//     is gone, so even if defaultFireFn's IsLive guard would no-op
+//     the delivery, the entry has no business surviving a close.
+//   - timers watching childID have childID pruned from t.watched
+//     (and t.idleBaseline). If t.watched becomes empty the timer is
+//     cancelled and deleted; we deliberately do NOT synthesise a
+//     fire here. The parent already received any legitimate idle
+//     transition before close_process — see allWatchedIdleLocked's
+//     "treat as satisfied" comment, which only applies to a
+//     concurrent re-evaluation, not to this explicit-removal hook.
+//
+// The natural-exit path (reapChild → emitExit for agent/command
+// kinds) is NOT routed through here: the classifier emits a final
+// idle transition on exit, which fires and deletes any watching
+// timers exactly once. Cancelling on exit would swallow that
+// legitimate fire and leave the parent never notified.
+func (m *timerManager) onChildClosed(childID string) {
+	m.mu.Lock()
+	changed := false
+	for id, t := range m.timers {
+		if t.ownerID == childID {
+			if t.rt != nil {
+				t.rt.Stop()
+				t.rt = nil
+			}
+			t.status = timerStatusCanceled
+			delete(m.timers, id)
+			changed = true
+			continue
+		}
+		if !contains(t.watched, childID) {
+			continue
+		}
+		pruned := t.watched[:0]
+		for _, w := range t.watched {
+			if w != childID {
+				pruned = append(pruned, w)
+			}
+		}
+		t.watched = pruned
+		if t.idleBaseline != nil {
+			delete(t.idleBaseline, childID)
+		}
+		changed = true
+		if len(t.watched) == 0 {
+			if t.rt != nil {
+				t.rt.Stop()
+				t.rt = nil
+			}
+			t.status = timerStatusCanceled
+			delete(m.timers, id)
+		}
+	}
+	m.mu.Unlock()
+	if changed {
+		m.notifyChanged()
+	}
+}
+
+// allWatchedIdleLocked reports whether every watched child is now
+// idle. Called with m.mu held — uses live Child.IdleState() under the
+// child's own atomic, not under m.mu.
+func (m *timerManager) allWatchedIdleLocked(t *pendingTimer) bool {
+	for _, id := range t.watched {
+		c := m.sess.FindChild(id)
+		if c == nil {
+			continue // disappeared; treat as satisfied so we don't hang
+		}
+		if !isIdleState(c.IdleState()) {
+			return false
+		}
+	}
+	return true
+}
+
+// TimerCancel removes a pending or paused timer owned by ownerID.
+func (m *timerManager) TimerCancel(ownerID, id string) error {
+	m.mu.Lock()
+	t, ok := m.timers[id]
+	if !ok {
+		m.mu.Unlock()
+		return mcp.Errorf(mcp.ErrorKindNotFound, "no such timer %q", id)
+	}
+	// Empty ownerID = top-level orchestrator caller (e.g. a non-agent
+	// MCP client); allow it to manage every timer in the session.
+	// Otherwise the caller's own id must match the timer's owner.
+	if ownerID != "" && t.ownerID != ownerID {
+		m.mu.Unlock()
+		return mcp.Errorf(mcp.ErrorKindRoleForbidden, "timer %q is not owned by caller", id)
+	}
+	if t.status == timerStatusFired || t.status == timerStatusCanceled {
+		// Cancelling a fired/cancelled timer is idempotent.
+		m.mu.Unlock()
+		return nil
+	}
+	if t.rt != nil {
+		t.rt.Stop()
+		t.rt = nil
+	}
+	t.status = timerStatusCanceled
+	delete(m.timers, id)
+	m.mu.Unlock()
+	m.notifyChanged()
+	return nil
+}
+
+// TimerPause stops the delay clock (or detaches the idle watch) but
+// keeps the timer in the registry.
+func (m *timerManager) TimerPause(ownerID, id string) error {
+	m.mu.Lock()
+	t, ok := m.timers[id]
+	if !ok {
+		m.mu.Unlock()
+		return mcp.Errorf(mcp.ErrorKindNotFound, "no such timer %q", id)
+	}
+	// Empty ownerID = top-level orchestrator caller (e.g. a non-agent
+	// MCP client); allow it to manage every timer in the session.
+	// Otherwise the caller's own id must match the timer's owner.
+	if ownerID != "" && t.ownerID != ownerID {
+		m.mu.Unlock()
+		return mcp.Errorf(mcp.ErrorKindRoleForbidden, "timer %q is not owned by caller", id)
+	}
+	if t.status != timerStatusPending {
+		m.mu.Unlock()
+		return mcp.Errorf(mcp.ErrorKindInvalidArgs, "timer %q is not pending", id)
+	}
+	if t.rt != nil {
+		t.pausedRemaining = time.Until(t.firesAt)
+		if t.pausedRemaining < 0 {
+			t.pausedRemaining = 0
+		}
+		t.rt.Stop()
+		t.rt = nil
+		// For idle_* timers, only the max-wait timer rides on rt — the
+		// idle-evaluation path lives in onChildStateChanged. Mark the
+		// pause so resume rearms the right thing.
+		t.pausedWasMaxWait = t.kind != timerKindDelay
+	}
+	t.status = timerStatusPaused
+	m.mu.Unlock()
+	m.notifyChanged()
+	return nil
+}
+
+// TimerResume re-arms a paused timer. For delay timers the remaining
+// duration is restored; idle-* timers re-attach to the state-change
+// watch list, and any remaining max-wait clock resumes.
+//
+// Idle-* timers also re-check their satisfaction condition immediately
+// on resume: idle transitions that occurred while paused are otherwise
+// missed (onChildStateChanged only sees future flips), so a child that
+// went idle during the pause window would never fire the timer. For
+// idle_any we look for any non-baseline watched child currently idle;
+// for idle_all we check whether every watched child is now idle.
+func (m *timerManager) TimerResume(ownerID, id string) error {
+	m.mu.Lock()
+	t, ok := m.timers[id]
+	if !ok {
+		m.mu.Unlock()
+		return mcp.Errorf(mcp.ErrorKindNotFound, "no such timer %q", id)
+	}
+	// Empty ownerID = top-level orchestrator caller (e.g. a non-agent
+	// MCP client); allow it to manage every timer in the session.
+	// Otherwise the caller's own id must match the timer's owner.
+	if ownerID != "" && t.ownerID != ownerID {
+		m.mu.Unlock()
+		return mcp.Errorf(mcp.ErrorKindRoleForbidden, "timer %q is not owned by caller", id)
+	}
+	if t.status != timerStatusPaused {
+		m.mu.Unlock()
+		return mcp.Errorf(mcp.ErrorKindInvalidArgs, "timer %q is not paused", id)
+	}
+	t.status = timerStatusPending
+	if t.pausedRemaining > 0 {
+		t.firesAt = time.Now().Add(t.pausedRemaining)
+		switch t.kind {
+		case timerKindDelay:
+			localID := id
+			t.rt = time.AfterFunc(t.pausedRemaining, func() { m.fireDelay(localID) })
+		default:
+			localID := id
+			t.rt = time.AfterFunc(t.pausedRemaining, func() { m.fireIdleMaxWait(localID) })
+		}
+		t.pausedRemaining = 0
+		t.pausedWasMaxWait = false
+	}
+	// For idle-* timers, evaluate the condition right now in case a
+	// watched child went idle while paused.
+	var fireNow bool
+	var owner *Child
+	var body, label string
+	switch t.kind {
+	case timerKindIdleAny:
+		for _, wid := range t.watched {
+			if t.idleBaseline[wid] {
+				continue
+			}
+			c := m.sess.FindChild(wid)
+			if c != nil && isIdleState(c.IdleState()) {
+				fireNow = true
+				break
+			}
+		}
+	case timerKindIdleAll:
+		if m.allWatchedIdleLocked(t) {
+			fireNow = true
+		}
+	}
+	if fireNow {
+		t.status = timerStatusFired
+		if t.rt != nil {
+			t.rt.Stop()
+			t.rt = nil
+		}
+		owner = m.sess.FindChild(t.ownerID)
+		body, label = t.body, t.label
+		delete(m.timers, id)
+	}
+	m.mu.Unlock()
+	m.notifyChanged()
+	if fireNow {
+		m.fireFn(owner, body, label)
+	}
+	return nil
+}
+
+// TimerList returns timers owned by ownerID, oldest-first. An empty
+// ownerID lists every active timer — the top-level orchestrator view.
+func (m *timerManager) TimerList(ownerID string) []mcp.TimerInfo {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	out := make([]mcp.TimerInfo, 0)
+	for _, t := range m.timers {
+		if ownerID != "" && t.ownerID != ownerID {
+			continue
+		}
+		if t.status != timerStatusPending && t.status != timerStatusPaused {
+			continue
+		}
+		info := mcp.TimerInfo{
+			ID:         t.id,
+			Label:      t.label,
+			Body:       t.body,
+			Kind:       string(t.kind),
+			Status:     t.status,
+			OwnerID:    t.ownerID,
+			WatchedIDs: append([]string(nil), t.watched...),
+		}
+		if t.status == timerStatusPending && !t.firesAt.IsZero() {
+			info.FiresAtUnixMS = t.firesAt.UnixMilli()
+		}
+		if t.status == timerStatusPaused && t.pausedRemaining > 0 {
+			info.PausedRemainingMS = t.pausedRemaining.Milliseconds()
+		}
+		out = append(out, info)
+	}
+	return out
+}
+
+// activeForChild returns the nearest pending or paused timer attached
+// to child id (either owned by it or watching it). Used by the sidebar
+// for the "⏱ 12s" indicator. nil when none.
+func (m *timerManager) activeForChild(id string) *mcp.TimerInfo {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	var best *pendingTimer
+	for _, t := range m.timers {
+		if t.status != timerStatusPending && t.status != timerStatusPaused {
+			continue
+		}
+		if t.ownerID != id && !contains(t.watched, id) {
+			continue
+		}
+		if best == nil {
+			best = t
+			continue
+		}
+		if t.firesAt.Before(best.firesAt) && !t.firesAt.IsZero() {
+			best = t
+		}
+	}
+	if best == nil {
+		return nil
+	}
+	info := mcp.TimerInfo{
+		ID:      best.id,
+		Label:   best.label,
+		Kind:    string(best.kind),
+		Status:  best.status,
+		OwnerID: best.ownerID,
+	}
+	if best.status == timerStatusPending && !best.firesAt.IsZero() {
+		info.FiresAtUnixMS = best.firesAt.UnixMilli()
+	}
+	if best.status == timerStatusPaused {
+		info.PausedRemainingMS = best.pausedRemaining.Milliseconds()
+	}
+	return &info
+}
+
+const (
+	timerSidebarMinRefresh       = 50 * time.Millisecond
+	timerSidebarSubsecondRefresh = 100 * time.Millisecond
+)
+
+func nextTimerSidebarLabelChange(d time.Duration) time.Duration {
+	if d <= 0 {
+		return 0
+	}
+	if d < time.Second {
+		if d < timerSidebarSubsecondRefresh {
+			return d
+		}
+		return timerSidebarSubsecondRefresh
+	}
+
+	step := time.Second
+	if d >= time.Hour {
+		step = time.Hour
+	} else if d >= time.Minute {
+		step = time.Minute
+	}
+	wait := d % step
+	if wait <= 0 || wait < timerSidebarMinRefresh {
+		return timerSidebarMinRefresh
+	}
+	return wait
+}
+
+func (m *timerManager) nextSidebarRefreshAfter(now time.Time) (time.Duration, bool) {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	var best time.Duration
+	found := false
+	for _, t := range m.timers {
+		if t.status != timerStatusPending || t.firesAt.IsZero() {
+			continue
+		}
+		wait := nextTimerSidebarLabelChange(t.firesAt.Sub(now))
+		if wait <= 0 {
+			wait = timerSidebarMinRefresh
+		}
+		if !found || wait < best {
+			best = wait
+			found = true
+		}
+	}
+	return best, found
+}
+
+func isIdleState(s IdleState) bool {
+	return s == StateIdle
+}
+
+func contains(haystack []string, needle string) bool {
+	for _, h := range haystack {
+		if h == needle {
+			return true
+		}
+	}
+	return false
+}

internal/app/timers_test.go

@@ -0,0 +1,675 @@
+package app
+
+import (
+	"sync"
+	"testing"
+	"time"
+)
+
+// recorderFire collects timer firings without touching a PTY. Lets the
+// timer manager run end-to-end logic in unit tests.
+type recorderFire struct {
+	mu    sync.Mutex
+	fires []recordedFire
+}
+
+type recordedFire struct {
+	OwnerID string
+	Body    string
+	Label   string
+}
+
+func (r *recorderFire) fn(owner *Child, body, label string) {
+	r.mu.Lock()
+	defer r.mu.Unlock()
+	id := ""
+	if owner != nil {
+		id = owner.ID
+	}
+	r.fires = append(r.fires, recordedFire{OwnerID: id, Body: body, Label: label})
+}
+
+func (r *recorderFire) snapshot() []recordedFire {
+	r.mu.Lock()
+	defer r.mu.Unlock()
+	out := make([]recordedFire, len(r.fires))
+	copy(out, r.fires)
+	return out
+}
+
+// fakeChild constructs a Child shell suitable for timer-manager tests.
+// Doesn't open a PTY — fireFn is overridden so InjectAsOrchestrator is
+// never reached.
+func fakeChild(id string) *Child {
+	c := newChildEntry(id, id, KindAgent, []string{"echo"}, nil, "", "", "")
+	running := StatusRunning
+	c.status.Store(&running)
+	return c
+}
+
+// addChild bypasses Spawn (no PTY needed) so the manager can find the
+// child by id and read its IdleState.
+func addChild(s *Session, c *Child) {
+	s.mu.Lock()
+	s.children[c.ID] = c
+	s.order = append(s.order, c.ID)
+	s.mu.Unlock()
+}
+
+func newTestManager(t *testing.T) (*Session, *timerManager, *recorderFire) {
+	t.Helper()
+	sess := NewSession(t.TempDir(), "test")
+	mgr := newTimerManager(sess)
+	rec := &recorderFire{}
+	mgr.fireFn = rec.fn
+	return sess, mgr, rec
+}
+
+func waitTimerChange(t *testing.T, mgr *timerManager) {
+	t.Helper()
+	select {
+	case <-mgr.changeEvents():
+	case <-time.After(time.Second):
+		t.Fatal("timed out waiting for timer change signal")
+	}
+}
+
+func TestNextTimerSidebarLabelChange(t *testing.T) {
+	tests := []struct {
+		name string
+		d    time.Duration
+		want time.Duration
+	}{
+		{name: "minutes", d: 2*time.Minute + 10*time.Second, want: 10 * time.Second},
+		{name: "minute_to_seconds", d: time.Minute + 500*time.Millisecond, want: 500 * time.Millisecond},
+		{name: "seconds", d: 59*time.Second + 500*time.Millisecond, want: 500 * time.Millisecond},
+		{name: "subsecond", d: 500 * time.Millisecond, want: timerSidebarSubsecondRefresh},
+		{name: "nearly_done", d: 30 * time.Millisecond, want: 30 * time.Millisecond},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			if got := nextTimerSidebarLabelChange(tt.d); got != tt.want {
+				t.Fatalf("nextTimerSidebarLabelChange(%s) = %s, want %s", tt.d, got, tt.want)
+			}
+		})
+	}
+}
+
+func TestTimerSidebarRefreshAfterUsesSoonestActiveBoundary(t *testing.T) {
+	_, mgr, _ := newTestManager(t)
+	now := time.Unix(123, 0)
+	mgr.mu.Lock()
+	mgr.timers["slow"] = &pendingTimer{
+		id:      "slow",
+		status:  timerStatusPending,
+		firesAt: now.Add(2*time.Minute + 10*time.Second),
+	}
+	mgr.timers["fast"] = &pendingTimer{
+		id:      "fast",
+		status:  timerStatusPending,
+		firesAt: now.Add(59*time.Second + 500*time.Millisecond),
+	}
+	mgr.timers["paused"] = &pendingTimer{
+		id:      "paused",
+		status:  timerStatusPaused,
+		firesAt: now.Add(100 * time.Millisecond),
+	}
+	mgr.mu.Unlock()
+
+	got, ok := mgr.nextSidebarRefreshAfter(now)
+	if !ok {
+		t.Fatal("nextSidebarRefreshAfter did not find active timers")
+	}
+	if got != 500*time.Millisecond {
+		t.Fatalf("nextSidebarRefreshAfter = %s, want 500ms", got)
+	}
+}
+
+func TestTimerManagerSignalsChangesForSidebar(t *testing.T) {
+	sess, mgr, _ := newTestManager(t)
+	owner := fakeChild("p_owner")
+	addChild(sess, owner)
+
+	id, err := mgr.TimerSet("p_owner", "x", "", 60)
+	if err != nil {
+		t.Fatalf("TimerSet: %v", err)
+	}
+	waitTimerChange(t, mgr)
+
+	if err := mgr.TimerPause("p_owner", id); err != nil {
+		t.Fatalf("TimerPause: %v", err)
+	}
+	waitTimerChange(t, mgr)
+
+	if err := mgr.TimerResume("p_owner", id); err != nil {
+		t.Fatalf("TimerResume: %v", err)
+	}
+	waitTimerChange(t, mgr)
+
+	if err := mgr.TimerCancel("p_owner", id); err != nil {
+		t.Fatalf("TimerCancel: %v", err)
+	}
+	waitTimerChange(t, mgr)
+}
+
+func TestTimerSetDelivers(t *testing.T) {
+	sess, mgr, rec := newTestManager(t)
+	c := fakeChild("p_owner")
+	addChild(sess, c)
+	id, err := mgr.TimerSet("p_owner", "wake up", "test", 0.05)
+	if err != nil {
+		t.Fatalf("TimerSet: %v", err)
+	}
+	if id == "" {
+		t.Fatal("empty timer id")
+	}
+	deadline := time.Now().Add(time.Second)
+	for time.Now().Before(deadline) {
+		if len(rec.snapshot()) > 0 {
+			break
+		}
+		time.Sleep(10 * time.Millisecond)
+	}
+	got := rec.snapshot()
+	if len(got) != 1 {
+		t.Fatalf("got %d fires, want 1", len(got))
+	}
+	if got[0].Body != "wake up" || got[0].OwnerID != "p_owner" {
+		t.Fatalf("unexpected fire: %+v", got[0])
+	}
+}
+
+func TestTimerIdleAllAlreadySatisfied(t *testing.T) {
+	sess, mgr, rec := newTestManager(t)
+	owner := fakeChild("p_owner")
+	a := fakeChild("p_a")
+	b := fakeChild("p_b")
+	addChild(sess, owner)
+	addChild(sess, a)
+	addChild(sess, b)
+	idle := StateIdle
+	a.idleState.Store(&idle)
+	b.idleState.Store(&idle)
+	resp, err := mgr.TimerFireWhenIdleAll("p_owner", "all done", "", []string{"p_a", "p_b"}, 0)
+	if err != nil {
+		t.Fatalf("TimerFireWhenIdleAll: %v", err)
+	}
+	if resp.Status != "already_satisfied" {
+		t.Fatalf("status: got %q want already_satisfied", resp.Status)
+	}
+	// fire is dispatched on a goroutine; wait briefly.
+	time.Sleep(50 * time.Millisecond)
+	got := rec.snapshot()
+	if len(got) != 1 || got[0].Body != "all done" {
+		t.Fatalf("fires: %+v", got)
+	}
+}
+
+func TestTimerIdleAnyFiresOnTransition(t *testing.T) {
+	sess, mgr, rec := newTestManager(t)
+	owner := fakeChild("p_owner")
+	a := fakeChild("p_a")
+	addChild(sess, owner)
+	addChild(sess, a)
+	// p_a starts busy.
+	working := StateWorking
+	a.idleState.Store(&working)
+	resp, err := mgr.TimerFireWhenIdleAny("p_owner", "one done", "", []string{"p_a"}, 0)
+	if err != nil {
+		t.Fatalf("TimerFireWhenIdleAny: %v", err)
+	}
+	if resp.Status != "pending" {
+		t.Fatalf("status: got %q want pending", resp.Status)
+	}
+	// Flip a into idle and deliver the state-change event.
+	idle := StateIdle
+	a.idleState.Store(&idle)
+	mgr.onChildStateChanged("p_a", StateIdle)
+	got := rec.snapshot()
+	if len(got) != 1 || got[0].Body != "one done" {
+		t.Fatalf("fires: %+v", got)
+	}
+}
+
+func TestTimerIdleAnyExcludesBaseline(t *testing.T) {
+	sess, mgr, rec := newTestManager(t)
+	owner := fakeChild("p_owner")
+	a := fakeChild("p_a")
+	addChild(sess, owner)
+	addChild(sess, a)
+	idle := StateIdle
+	a.idleState.Store(&idle)
+	resp, err := mgr.TimerFireWhenIdleAny("p_owner", "one done", "", []string{"p_a"}, 0)
+	if err != nil {
+		t.Fatalf("TimerFireWhenIdleAny: %v", err)
+	}
+	if resp.Status != "pending" {
+		t.Fatalf("status: got %q want pending", resp.Status)
+	}
+	// Send a redundant idle transition for p_a; should NOT fire because
+	// p_a was idle at registration.
+	mgr.onChildStateChanged("p_a", StateIdle)
+	if got := rec.snapshot(); len(got) != 0 {
+		t.Fatalf("unexpected fires: %+v", got)
+	}
+}
+
+func TestTimerCancelPauseResume(t *testing.T) {
+	sess, mgr, rec := newTestManager(t)
+	owner := fakeChild("p_owner")
+	addChild(sess, owner)
+
+	// Cancel before fire.
+	id, _ := mgr.TimerSet("p_owner", "x", "", 0.2)
+	if err := mgr.TimerCancel("p_owner", id); err != nil {
+		t.Fatalf("Cancel: %v", err)
+	}
+	time.Sleep(300 * time.Millisecond)
+	if got := rec.snapshot(); len(got) != 0 {
+		t.Fatalf("cancel didn't stop fire: %+v", got)
+	}
+
+	// Pause then resume → fire after resume.
+	id2, _ := mgr.TimerSet("p_owner", "y", "", 0.2)
+	time.Sleep(50 * time.Millisecond)
+	if err := mgr.TimerPause("p_owner", id2); err != nil {
+		t.Fatalf("Pause: %v", err)
+	}
+	time.Sleep(300 * time.Millisecond) // would have fired by now if not paused
+	if got := rec.snapshot(); len(got) != 0 {
+		t.Fatalf("paused timer fired: %+v", got)
+	}
+	if err := mgr.TimerResume("p_owner", id2); err != nil {
+		t.Fatalf("Resume: %v", err)
+	}
+	deadline := time.Now().Add(time.Second)
+	for time.Now().Before(deadline) {
+		if len(rec.snapshot()) > 0 {
+			break
+		}
+		time.Sleep(20 * time.Millisecond)
+	}
+	if got := rec.snapshot(); len(got) != 1 || got[0].Body != "y" {
+		t.Fatalf("resume fire: %+v", got)
+	}
+}
+
+func TestTimerOwnershipEnforced(t *testing.T) {
+	sess, mgr, _ := newTestManager(t)
+	a := fakeChild("p_a")
+	b := fakeChild("p_b")
+	addChild(sess, a)
+	addChild(sess, b)
+	id, _ := mgr.TimerSet("p_a", "hi", "", 60)
+	if err := mgr.TimerCancel("p_b", id); err == nil {
+		t.Fatal("expected ownership error from foreign cancel")
+	}
+	if err := mgr.TimerPause("p_b", id); err == nil {
+		t.Fatal("expected ownership error from foreign pause")
+	}
+}
+
+// TestTimerResumeRechecksIdleAll covers the case where every watched
+// child becomes idle while an idle_all timer is paused. Without a resume
+// re-check, the timer would stay pending forever because the state
+// transitions happened during the pause window.
+func TestTimerResumeRechecksIdleAll(t *testing.T) {
+	sess, mgr, rec := newTestManager(t)
+	owner := fakeChild("p_owner")
+	a := fakeChild("p_a")
+	b := fakeChild("p_b")
+	addChild(sess, owner)
+	addChild(sess, a)
+	addChild(sess, b)
+	working := StateWorking
+	a.idleState.Store(&working)
+	b.idleState.Store(&working)
+
+	resp, err := mgr.TimerFireWhenIdleAll("p_owner", "all done", "", []string{"p_a", "p_b"}, 0)
+	if err != nil {
+		t.Fatalf("TimerFireWhenIdleAll: %v", err)
+	}
+	if resp.Status != "pending" {
+		t.Fatalf("status: got %q want pending", resp.Status)
+	}
+	if err := mgr.TimerPause("p_owner", resp.ID); err != nil {
+		t.Fatalf("Pause: %v", err)
+	}
+
+	// Both watched children become idle WHILE THE TIMER IS PAUSED, so
+	// onChildStateChanged is not consulted for this timer.
+	idle := StateIdle
+	a.idleState.Store(&idle)
+	b.idleState.Store(&idle)
+
+	if err := mgr.TimerResume("p_owner", resp.ID); err != nil {
+		t.Fatalf("Resume: %v", err)
+	}
+	got := rec.snapshot()
+	if len(got) != 1 || got[0].Body != "all done" {
+		t.Fatalf("expected fire on resume, got: %+v", got)
+	}
+}
+
+// TestTimerResumeRechecksIdleAny covers the same missed-transition shape
+// for idle_any: a non-baseline watched child going idle during pause must
+// fire on resume.
+func TestTimerResumeRechecksIdleAny(t *testing.T) {
+	sess, mgr, rec := newTestManager(t)
+	owner := fakeChild("p_owner")
+	a := fakeChild("p_a")
+	addChild(sess, owner)
+	addChild(sess, a)
+	working := StateWorking
+	a.idleState.Store(&working)
+
+	resp, err := mgr.TimerFireWhenIdleAny("p_owner", "one done", "", []string{"p_a"}, 0)
+	if err != nil {
+		t.Fatalf("TimerFireWhenIdleAny: %v", err)
+	}
+	if resp.Status != "pending" {
+		t.Fatalf("status: got %q want pending", resp.Status)
+	}
+	if err := mgr.TimerPause("p_owner", resp.ID); err != nil {
+		t.Fatalf("Pause: %v", err)
+	}
+	idle := StateIdle
+	a.idleState.Store(&idle)
+	if err := mgr.TimerResume("p_owner", resp.ID); err != nil {
+		t.Fatalf("Resume: %v", err)
+	}
+	got := rec.snapshot()
+	if len(got) != 1 || got[0].Body != "one done" {
+		t.Fatalf("expected fire on resume, got: %+v", got)
+	}
+}
+
+// TestTimerResumeIdleAnyExcludesBaselineDuringPause guards against a
+// resume re-check firing for a watcher that was idle at registration
+// (and therefore part of the baseline) — only non-baseline transitions
+// should satisfy idle_any.
+func TestTimerResumeIdleAnyExcludesBaselineDuringPause(t *testing.T) {
+	sess, mgr, rec := newTestManager(t)
+	owner := fakeChild("p_owner")
+	a := fakeChild("p_a")
+	b := fakeChild("p_b")
+	addChild(sess, owner)
+	addChild(sess, a)
+	addChild(sess, b)
+	idle := StateIdle
+	working := StateWorking
+	a.idleState.Store(&idle)    // baseline: already idle
+	b.idleState.Store(&working) // not baseline
+
+	resp, err := mgr.TimerFireWhenIdleAny("p_owner", "one done", "", []string{"p_a", "p_b"}, 0)
+	if err != nil {
+		t.Fatalf("TimerFireWhenIdleAny: %v", err)
+	}
+	if err := mgr.TimerPause("p_owner", resp.ID); err != nil {
+		t.Fatalf("Pause: %v", err)
+	}
+	// b stays working — only a is idle, and a was baseline. Resume
+	// must not fire.
+	if err := mgr.TimerResume("p_owner", resp.ID); err != nil {
+		t.Fatalf("Resume: %v", err)
+	}
+	if got := rec.snapshot(); len(got) != 0 {
+		t.Fatalf("unexpected fire on resume: %+v", got)
+	}
+}
+
+// TestTimerRecordsRemovedOnFire ensures fired delay timers don't leak
+// in the timer registry — bodies and metadata must be released.
+func TestTimerRecordsRemovedOnFire(t *testing.T) {
+	sess, mgr, rec := newTestManager(t)
+	c := fakeChild("p_owner")
+	addChild(sess, c)
+	id, err := mgr.TimerSet("p_owner", "wake up", "test", 0.05)
+	if err != nil {
+		t.Fatalf("TimerSet: %v", err)
+	}
+	deadline := time.Now().Add(time.Second)
+	for time.Now().Before(deadline) {
+		if len(rec.snapshot()) > 0 {
+			break
+		}
+		time.Sleep(10 * time.Millisecond)
+	}
+	if len(rec.snapshot()) != 1 {
+		t.Fatalf("timer didn't fire")
+	}
+	mgr.mu.Lock()
+	_, stillThere := mgr.timers[id]
+	count := len(mgr.timers)
+	mgr.mu.Unlock()
+	if stillThere {
+		t.Fatalf("fired timer %s was not removed from registry", id)
+	}
+	if count != 0 {
+		t.Fatalf("timer registry not drained: %d entries", count)
+	}
+}
+
+// TestTimerRecordsRemovedOnCancel ensures canceled timers are dropped
+// from the registry.
+func TestTimerRecordsRemovedOnCancel(t *testing.T) {
+	sess, mgr, _ := newTestManager(t)
+	c := fakeChild("p_owner")
+	addChild(sess, c)
+	id, err := mgr.TimerSet("p_owner", "x", "", 60)
+	if err != nil {
+		t.Fatalf("TimerSet: %v", err)
+	}
+	if err := mgr.TimerCancel("p_owner", id); err != nil {
+		t.Fatalf("Cancel: %v", err)
+	}
+	mgr.mu.Lock()
+	_, stillThere := mgr.timers[id]
+	mgr.mu.Unlock()
+	if stillThere {
+		t.Fatalf("canceled timer %s was not removed from registry", id)
+	}
+}
+
+// TestTimerRecordsRemovedOnIdleFire ensures idle_* timers are dropped
+// from the registry once they fire via onChildStateChanged.
+func TestTimerRecordsRemovedOnIdleFire(t *testing.T) {
+	sess, mgr, rec := newTestManager(t)
+	owner := fakeChild("p_owner")
+	a := fakeChild("p_a")
+	addChild(sess, owner)
+	addChild(sess, a)
+	working := StateWorking
+	a.idleState.Store(&working)
+	resp, err := mgr.TimerFireWhenIdleAny("p_owner", "one done", "", []string{"p_a"}, 0)
+	if err != nil {
+		t.Fatalf("TimerFireWhenIdleAny: %v", err)
+	}
+	idle := StateIdle
+	a.idleState.Store(&idle)
+	mgr.onChildStateChanged("p_a", StateIdle)
+	if got := rec.snapshot(); len(got) != 1 {
+		t.Fatalf("expected fire, got: %+v", got)
+	}
+	mgr.mu.Lock()
+	_, stillThere := mgr.timers[resp.ID]
+	mgr.mu.Unlock()
+	if stillThere {
+		t.Fatalf("fired idle timer %s was not removed from registry", resp.ID)
+	}
+}
+
+// TestTimerCloseChildPrunesWatched covers the happy partial-prune
+// case: an idle_any timer watches two children, one is closed, the
+// timer stays pending and the remaining child can still satisfy it.
+func TestTimerCloseChildPrunesWatched(t *testing.T) {
+	sess, mgr, rec := newTestManager(t)
+	owner := fakeChild("p_owner")
+	a := fakeChild("p_a")
+	b := fakeChild("p_b")
+	addChild(sess, owner)
+	addChild(sess, a)
+	addChild(sess, b)
+	working := StateWorking
+	a.idleState.Store(&working)
+	b.idleState.Store(&working)
+
+	resp, err := mgr.TimerFireWhenIdleAny("p_owner", "one done", "", []string{"p_a", "p_b"}, 0)
+	if err != nil {
+		t.Fatalf("TimerFireWhenIdleAny: %v", err)
+	}
+
+	mgr.onChildClosed("p_a")
+
+	mgr.mu.Lock()
+	t1, ok := mgr.timers[resp.ID]
+	if !ok {
+		mgr.mu.Unlock()
+		t.Fatalf("timer was removed but still has live watched")
+	}
+	watched := append([]string(nil), t1.watched...)
+	mgr.mu.Unlock()
+	if len(watched) != 1 || watched[0] != "p_b" {
+		t.Fatalf("watched after close: %v, want [p_b]", watched)
+	}
+	if got := rec.snapshot(); len(got) != 0 {
+		t.Fatalf("close synthesised a fire: %+v", got)
+	}
+
+	// p_b can still satisfy the timer.
+	idle := StateIdle
+	b.idleState.Store(&idle)
+	mgr.onChildStateChanged("p_b", StateIdle)
+	if got := rec.snapshot(); len(got) != 1 || got[0].Body != "one done" {
+		t.Fatalf("post-prune fire: %+v", got)
+	}
+}
+
+// TestTimerCloseLastWatchedCancels is the regression for the
+// reported stale-fire symptom: the only watched child is closed,
+// so the timer must be cancelled — no synthetic fire, and the
+// registry entry must be gone so a trailing classifier tick for the
+// removed child cannot re-deliver later.
+func TestTimerCloseLastWatchedCancels(t *testing.T) {
+	sess, mgr, rec := newTestManager(t)
+	owner := fakeChild("p_owner")
+	a := fakeChild("p_a")
+	addChild(sess, owner)
+	addChild(sess, a)
+	working := StateWorking
+	a.idleState.Store(&working)
+
+	resp, err := mgr.TimerFireWhenIdleAny("p_owner", "stale body", "", []string{"p_a"}, 0)
+	if err != nil {
+		t.Fatalf("TimerFireWhenIdleAny: %v", err)
+	}
+
+	mgr.onChildClosed("p_a")
+
+	mgr.mu.Lock()
+	_, stillThere := mgr.timers[resp.ID]
+	mgr.mu.Unlock()
+	if stillThere {
+		t.Fatalf("timer with no remaining watched should be removed")
+	}
+	if got := rec.snapshot(); len(got) != 0 {
+		t.Fatalf("close synthesised a fire: %+v", got)
+	}
+
+	// Simulate the trailing classifier tick for the now-closed child —
+	// must not fire.
+	mgr.onChildStateChanged("p_a", StateIdle)
+	if got := rec.snapshot(); len(got) != 0 {
+		t.Fatalf("trailing state change re-fired: %+v", got)
+	}
+}
+
+// TestTimerCloseChildIdleAllPartialPrune mirrors the idle_any
+// partial-prune for idle_all: pruning a watched child shrinks the
+// list; the remaining child going idle then satisfies the timer.
+func TestTimerCloseChildIdleAllPartialPrune(t *testing.T) {
+	sess, mgr, rec := newTestManager(t)
+	owner := fakeChild("p_owner")
+	a := fakeChild("p_a")
+	b := fakeChild("p_b")
+	addChild(sess, owner)
+	addChild(sess, a)
+	addChild(sess, b)
+	working := StateWorking
+	a.idleState.Store(&working)
+	b.idleState.Store(&working)
+
+	resp, err := mgr.TimerFireWhenIdleAll("p_owner", "all done", "", []string{"p_a", "p_b"}, 0)
+	if err != nil {
+		t.Fatalf("TimerFireWhenIdleAll: %v", err)
+	}
+	if resp.Status != "pending" {
+		t.Fatalf("status: got %q want pending", resp.Status)
+	}
+
+	mgr.onChildClosed("p_a")
+
+	idle := StateIdle
+	b.idleState.Store(&idle)
+	mgr.onChildStateChanged("p_b", StateIdle)
+	if got := rec.snapshot(); len(got) != 1 || got[0].Body != "all done" {
+		t.Fatalf("idle_all after partial prune: %+v", got)
+	}
+}
+
+// TestTimerCloseOwnerCancelsDelay ensures a delay timer is dropped
+// when its owner is closed: no delivery, registry empty, the
+// underlying time.Timer is stopped.
+func TestTimerCloseOwnerCancelsDelay(t *testing.T) {
+	sess, mgr, rec := newTestManager(t)
+	c := fakeChild("p_owner")
+	addChild(sess, c)
+	id, err := mgr.TimerSet("p_owner", "x", "", 0.1)
+	if err != nil {
+		t.Fatalf("TimerSet: %v", err)
+	}
+
+	mgr.onChildClosed("p_owner")
+
+	mgr.mu.Lock()
+	_, stillThere := mgr.timers[id]
+	mgr.mu.Unlock()
+	if stillThere {
+		t.Fatalf("delay timer was not removed when owner closed")
+	}
+	time.Sleep(200 * time.Millisecond) // past the original firesAt
+	if got := rec.snapshot(); len(got) != 0 {
+		t.Fatalf("delay timer fired after owner close: %+v", got)
+	}
+}
+
+// TestTimerCloseWatchedSubAgent is the exact shape of the reported
+// stale-fire bug: orchestrator registers a watcher on a sub-agent,
+// the sub-agent is closed, and the orchestrator must receive
+// nothing (no stale body delivered after close_process).
+func TestTimerCloseWatchedSubAgent(t *testing.T) {
+	sess, mgr, rec := newTestManager(t)
+	parent := fakeChild("p_owner")
+	sub := fakeChild("p_sub")
+	addChild(sess, parent)
+	addChild(sess, sub)
+	working := StateWorking
+	sub.idleState.Store(&working)
+
+	if _, err := mgr.TimerFireWhenIdleAny(
+		"p_owner",
+		"codex-review-591 finished. Read your own pane …",
+		"", []string{"p_sub"}, 0,
+	); err != nil {
+		t.Fatalf("TimerFireWhenIdleAny: %v", err)
+	}
+
+	mgr.onChildClosed("p_sub")
+
+	// Trailing classifier emission for the closed sub-agent must
+	// not deliver anything to the parent.
+	mgr.onChildStateChanged("p_sub", StateIdle)
+	if got := rec.snapshot(); len(got) != 0 {
+		t.Fatalf("stale fire delivered to parent after sub-agent close: %+v", got)
+	}
+}

internal/app/toast.go

@@ -0,0 +1,361 @@
+package app
+
+import (
+	"os"
+	"strings"
+	"sync"
+)
+
+// toastKind classifies a toast for styling and for migrating the
+// pre-existing flashError / flashTransient / notifyAttention call
+// sites onto the new stack.
+type toastKind int
+
+const (
+	toastInfo toastKind = iota
+	toastError
+	toastAttention
+)
+
+// toast is one entry in the host-level notification stack. Toasts
+// persist until the user dismisses them with Ctrl-N or the
+// "Clear notifications" palette command — there's no auto-expiry.
+type toast struct {
+	id   uint64
+	kind toastKind
+	text string
+}
+
+// toastStackCap caps how many toasts can be visible at once.
+// Older entries drop off the bottom when a new push would exceed it.
+const toastStackCap = 5
+
+// toastBoxMaxWidth bounds the rendered box width so a wide pane
+// doesn't produce huge toasts. Boxes shrink below this when the pane
+// is narrow.
+const toastBoxMaxWidth = 50
+
+// toastBoxMinWidth is the floor below which we refuse to render —
+// any narrower and there's not enough room for borders + content.
+const toastBoxMinWidth = 20
+
+// toastContentRows is how many lines of message body each toast box
+// reserves. The dismiss hint lives on the host status strip, so the
+// box itself is purely the message.
+const toastContentRows = 3
+
+// toastStack owns the ordered list of live toasts. Oldest at
+// index 0, newest (visually topmost) at the end. The stack's own
+// mutex is intentionally separate from uiState.mu so push / dismiss
+// can be called from any goroutine without participating in the
+// host's bigger lock-ordering rules.
+type toastStack struct {
+	mu    sync.Mutex
+	items []toast
+	next  uint64
+}
+
+func (s *toastStack) push(kind toastKind, text string) {
+	s.mu.Lock()
+	defer s.mu.Unlock()
+	s.next++
+	s.items = append(s.items, toast{id: s.next, kind: kind, text: text})
+	if len(s.items) > toastStackCap {
+		s.items = s.items[len(s.items)-toastStackCap:]
+	}
+}
+
+// dismissTop pops the most recent toast (the one rendered at the
+// top of the stack). Returns true if something was removed so
+// callers can decide whether to repaint.
+func (s *toastStack) dismissTop() bool {
+	s.mu.Lock()
+	defer s.mu.Unlock()
+	if len(s.items) == 0 {
+		return false
+	}
+	s.items = s.items[:len(s.items)-1]
+	return true
+}
+
+func (s *toastStack) clear() bool {
+	s.mu.Lock()
+	defer s.mu.Unlock()
+	if len(s.items) == 0 {
+		return false
+	}
+	s.items = s.items[:0]
+	return true
+}
+
+func (s *toastStack) snapshot() []toast {
+	s.mu.Lock()
+	defer s.mu.Unlock()
+	if len(s.items) == 0 {
+		return nil
+	}
+	out := make([]toast, len(s.items))
+	copy(out, s.items)
+	return out
+}
+
+func (s *toastStack) length() int {
+	s.mu.Lock()
+	defer s.mu.Unlock()
+	return len(s.items)
+}
+
+// notifyToast is the single entry point that the former flash
+// helpers now delegate to. It pushes onto the stack and triggers a
+// repaint of the focused surface so the new toast appears
+// immediately; the repaint path also re-renders the stack on top.
+func (st *uiState) notifyToast(kind toastKind, text string) {
+	st.toasts.push(kind, text)
+	st.refreshToastSurface()
+}
+
+// refreshToastSurface re-renders whatever surface the toasts are
+// drawn over (focused child, focused pad, or the empty-state
+// canvas). Each of those paths calls renderToasts at the end, so
+// the toast layer is always reapplied on top of a freshly-drawn
+// pane. Centralised so push / dismiss / clear share one code path.
+//
+// The status strip also gains/loses the "Ctrl-N · dismiss" hint as
+// the stack toggles between empty and non-empty, so we redraw it
+// here too rather than waiting for the chrome ticker.
+func (st *uiState) refreshToastSurface() {
+	st.mu.Lock()
+	focusedPad := st.focusedPad
+	focusedID := st.focusedID
+	palOpen := st.palette != nil
+	st.mu.Unlock()
+	if palOpen {
+		// Palette owns the whole screen while it's open; toasts will
+		// repaint via closePalette's restore path.
+		return
+	}
+	switch {
+	case focusedPad != "":
+		st.repaintFocusedPad()
+	case focusedID != "":
+		st.repaintFocused()
+	default:
+		st.renderEmptyState()
+	}
+	st.drawStatusLine()
+}
+
+// renderToasts draws the toast stack over the top-right of the
+// focused pane. Called from repaintFocused / repaintFocusedPad /
+// renderEmptyState after they finish so toasts always sit on top of
+// freshly-redrawn pane content. Safe to call when the stack is
+// empty (no-op).
+func (st *uiState) renderToasts() {
+	bytes := st.toastOverlayBytes()
+	if len(bytes) == 0 {
+		return
+	}
+	st.outMu.Lock()
+	defer st.outMu.Unlock()
+	_, _ = os.Stdout.Write(bytes)
+}
+
+// toastOverlayBytes builds the toast layer as a single byte buffer
+// without writing to stdout. Returns nil when the stack is empty or
+// the layout can't accommodate a box. Callers either write it
+// directly (renderToasts) or stitch it onto the end of another
+// stdout write so claude/codex/opencode redraws that paint over the
+// top-right region can't leave the toast half-erased.
+func (st *uiState) toastOverlayBytes() []byte {
+	items := st.toasts.snapshot()
+	if len(items) == 0 {
+		return nil
+	}
+	st.mu.Lock()
+	palOpen := st.palette != nil
+	st.mu.Unlock()
+	if palOpen {
+		return nil
+	}
+	layout := st.layoutSnapshot()
+	paneCols := int(layout.childCols())
+	paneRows := int(layout.childRows())
+	if paneCols < toastBoxMinWidth+2 || paneRows < toastContentRows+2 {
+		return nil
+	}
+	boxWidth := toastBoxMaxWidth
+	if max := paneCols - 4; max < boxWidth {
+		boxWidth = max
+	}
+	if boxWidth < toastBoxMinWidth {
+		return nil
+	}
+	contentWidth := boxWidth - 4 // 2 border cells + 2 inner padding
+	// Reserve two columns for the icon prefix on row 1 so wrapped rows
+	// indent under the body text rather than under the glyph.
+	const iconCols = 2
+	bodyRoom := contentWidth - iconCols
+	if bodyRoom < 1 {
+		return nil
+	}
+
+	var b strings.Builder
+	// Wrap the whole overlay in DECSET 2026 (synchronized output)
+	// brackets so terminals that support BSU/ESU buffer the box paint
+	// into a single frame — without this, claude's continuous redraws
+	// and our overlay race on each cell, producing visible flicker.
+	// Terminals that don't recognise 2026 ignore the brackets, so the
+	// fallback behaviour is the same as before.
+	b.WriteString("\x1b[?2026h\x1b7\x1b[?25l")
+
+	row := int(layout.mainTop) + 1
+	col := int(layout.mainLeft) + paneCols - boxWidth - 1
+	if col < int(layout.mainLeft) {
+		col = int(layout.mainLeft)
+	}
+
+	// Render newest first (visually on top), iterating items in
+	// reverse so the most recent push lands at the smallest row.
+	for idx := len(items) - 1; idx >= 0; idx-- {
+		t := items[idx]
+		height := toastContentRows + 2
+		// Stop if we'd run off the bottom of the pane.
+		if row+height > int(layout.mainTop)+paneRows {
+			break
+		}
+		border := toastBorderStyle(t.kind)
+		wrapped := wrapToastBody(t.text, bodyRoom)
+
+		// Top border.
+		moveTo(&b, row, col)
+		b.WriteString(border)
+		b.WriteString("╭")
+		b.WriteString(strings.Repeat("─", boxWidth-2))
+		b.WriteString("╮")
+		b.WriteString(styleReset)
+		row++
+
+		// Content rows. Row 0 carries the kind glyph; rows 1..N indent
+		// by iconCols spaces so wrapped text lines up under the body.
+		for i := 0; i < toastContentRows; i++ {
+			moveTo(&b, row, col)
+			b.WriteString(border)
+			b.WriteString("│")
+			b.WriteString(styleReset)
+			b.WriteString(" ")
+			if i == 0 {
+				b.WriteString(toastIcon(t.kind))
+			} else {
+				b.WriteString(strings.Repeat(" ", iconCols))
+			}
+			line := wrapped[i]
+			b.WriteString(line)
+			b.WriteString(strings.Repeat(" ", max(0, bodyRoom-visibleLen(line))))
+			b.WriteString(" ")
+			b.WriteString(border)
+			b.WriteString("│")
+			b.WriteString(styleReset)
+			row++
+		}
+
+		// Bottom border.
+		moveTo(&b, row, col)
+		b.WriteString(border)
+		b.WriteString("╰")
+		b.WriteString(strings.Repeat("─", boxWidth-2))
+		b.WriteString("╯")
+		b.WriteString(styleReset)
+		row++
+
+		// 1-row gap between stacked toasts.
+		row++
+	}
+
+	b.WriteString("\x1b[?25h\x1b8\x1b[?2026l")
+	return []byte(b.String())
+}
+
+func toastBorderStyle(kind toastKind) string {
+	switch kind {
+	case toastError:
+		return styleError
+	case toastAttention:
+		return styleAccent
+	default:
+		return styleBorder
+	}
+}
+
+// wrapToastBody word-wraps text into exactly toastContentRows lines,
+// each at most width visible runes wide. Short messages are padded
+// with empty trailing lines so callers can iterate a fixed-size
+// slice; messages that don't fit get ellipsized on the last line.
+func wrapToastBody(text string, width int) []string {
+	out := make([]string, toastContentRows)
+	if width < 1 {
+		return out
+	}
+	all := wrapToastWords(text, width)
+	if len(all) > toastContentRows {
+		all = all[:toastContentRows]
+		last := all[len(all)-1]
+		if visibleLen(last) >= width {
+			last = clipRunes(last, width-1) + "…"
+		} else {
+			last = last + "…"
+		}
+		all[len(all)-1] = last
+	}
+	for i, l := range all {
+		out[i] = l
+	}
+	return out
+}
+
+// wrapToastWords is a small word-wrapper sized for toast bodies:
+// greedy, breaks overlong words on rune boundaries, drops collapsing
+// whitespace via strings.Fields.
+func wrapToastWords(text string, width int) []string {
+	var lines []string
+	var cur string
+	flush := func() {
+		if cur != "" {
+			lines = append(lines, cur)
+			cur = ""
+		}
+	}
+	for _, word := range strings.Fields(text) {
+		for visibleLen(word) > width {
+			flush()
+			head := clipRunes(word, width)
+			lines = append(lines, head)
+			word = word[len(head):]
+		}
+		if word == "" {
+			continue
+		}
+		if cur == "" {
+			cur = word
+			continue
+		}
+		if visibleLen(cur)+1+visibleLen(word) <= width {
+			cur += " " + word
+			continue
+		}
+		flush()
+		cur = word
+	}
+	flush()
+	return lines
+}
+
+func toastIcon(kind toastKind) string {
+	switch kind {
+	case toastError:
+		return styleError + "✗ " + styleReset
+	case toastAttention:
+		return styleAccent + "! " + styleReset
+	default:
+		return styleHint + "• " + styleReset
+	}
+}

internal/app/toast_test.go

@@ -0,0 +1,164 @@
+package app
+
+import (
+	"strings"
+	"testing"
+)
+
+func TestToastStackPushAndOrder(t *testing.T) {
+	var s toastStack
+	s.push(toastInfo, "one")
+	s.push(toastError, "two")
+	s.push(toastAttention, "three")
+
+	snap := s.snapshot()
+	if len(snap) != 3 {
+		t.Fatalf("snapshot len = %d, want 3", len(snap))
+	}
+	if snap[0].text != "one" || snap[1].text != "two" || snap[2].text != "three" {
+		t.Fatalf("snapshot order wrong: %#v", snap)
+	}
+	if snap[0].kind != toastInfo || snap[1].kind != toastError || snap[2].kind != toastAttention {
+		t.Fatalf("snapshot kinds wrong: %#v", snap)
+	}
+	// IDs strictly increase.
+	if !(snap[0].id < snap[1].id && snap[1].id < snap[2].id) {
+		t.Fatalf("ids not increasing: %#v", snap)
+	}
+}
+
+func TestToastStackCapDropsOldest(t *testing.T) {
+	var s toastStack
+	for i := 0; i < toastStackCap+3; i++ {
+		s.push(toastInfo, "msg")
+	}
+	snap := s.snapshot()
+	if len(snap) != toastStackCap {
+		t.Fatalf("len = %d, want %d", len(snap), toastStackCap)
+	}
+	// The earliest IDs should have been dropped, leaving the highest
+	// toastStackCap IDs.
+	for i := 1; i < len(snap); i++ {
+		if snap[i].id <= snap[i-1].id {
+			t.Fatalf("ordering broken after cap: %#v", snap)
+		}
+	}
+	// First retained id should be 4 (1,2,3 dropped; cap=5 leaves 4..8).
+	want := uint64(toastStackCap + 3 - toastStackCap + 1)
+	if snap[0].id != want {
+		t.Fatalf("first retained id = %d, want %d", snap[0].id, want)
+	}
+}
+
+func TestToastStackDismissTop(t *testing.T) {
+	var s toastStack
+	if s.dismissTop() {
+		t.Fatalf("dismissTop on empty stack returned true")
+	}
+	s.push(toastInfo, "a")
+	s.push(toastError, "b")
+	if !s.dismissTop() {
+		t.Fatalf("dismissTop returned false with items present")
+	}
+	snap := s.snapshot()
+	if len(snap) != 1 || snap[0].text != "a" {
+		t.Fatalf("after dismissTop: %#v", snap)
+	}
+	if !s.dismissTop() {
+		t.Fatalf("dismissTop on last item returned false")
+	}
+	if s.length() != 0 {
+		t.Fatalf("length after final dismiss = %d, want 0", s.length())
+	}
+}
+
+func TestToastStackClear(t *testing.T) {
+	var s toastStack
+	if s.clear() {
+		t.Fatalf("clear on empty returned true")
+	}
+	s.push(toastInfo, "a")
+	s.push(toastError, "b")
+	s.push(toastAttention, "c")
+	if !s.clear() {
+		t.Fatalf("clear returned false with items present")
+	}
+	if s.length() != 0 {
+		t.Fatalf("length after clear = %d, want 0", s.length())
+	}
+	if snap := s.snapshot(); snap != nil {
+		t.Fatalf("snapshot after clear = %#v, want nil", snap)
+	}
+}
+
+func TestToastStackSnapshotIsCopy(t *testing.T) {
+	var s toastStack
+	s.push(toastInfo, "a")
+	snap := s.snapshot()
+	snap[0].text = "mutated"
+	again := s.snapshot()
+	if again[0].text != "a" {
+		t.Fatalf("snapshot is not an independent copy: %#v", again)
+	}
+}
+
+func TestWrapToastBodyFixedHeight(t *testing.T) {
+	got := wrapToastBody("short", 20)
+	if len(got) != toastContentRows {
+		t.Fatalf("len = %d, want %d", len(got), toastContentRows)
+	}
+	if got[0] != "short" {
+		t.Fatalf("line 0 = %q, want \"short\"", got[0])
+	}
+	if got[1] != "" || got[2] != "" {
+		t.Fatalf("trailing pads not empty: %#v", got)
+	}
+}
+
+func TestWrapToastBodyWrapsOnWordBoundary(t *testing.T) {
+	got := wrapToastBody("the quick brown fox jumps over", 10)
+	// Expect greedy fill: "the quick" (9), "brown fox" (9), "jumps over" (10).
+	want := []string{"the quick", "brown fox", "jumps over"}
+	for i, w := range want {
+		if got[i] != w {
+			t.Fatalf("line %d = %q, want %q (full=%#v)", i, got[i], w, got)
+		}
+	}
+}
+
+func TestWrapToastBodyEllipsizesOverflow(t *testing.T) {
+	got := wrapToastBody("alpha beta gamma delta epsilon zeta eta theta", 6)
+	if len(got) != toastContentRows {
+		t.Fatalf("len = %d, want %d", len(got), toastContentRows)
+	}
+	last := got[toastContentRows-1]
+	if !strings.HasSuffix(last, "…") {
+		t.Fatalf("overflow should ellipsize last line, got %q (full=%#v)", last, got)
+	}
+	if visibleLen(last) > 6 {
+		t.Fatalf("last line %q exceeds width 6", last)
+	}
+}
+
+func TestWrapToastBodyBreaksOverlongWord(t *testing.T) {
+	got := wrapToastBody("supercalifragilistic", 6)
+	if got[0] != "superc" {
+		t.Fatalf("line 0 = %q, want \"superc\"", got[0])
+	}
+	if got[1] != "alifra" {
+		t.Fatalf("line 1 = %q, want \"alifra\"", got[1])
+	}
+	// Third line should hold the rest (possibly ellipsized).
+	if got[2] == "" {
+		t.Fatalf("line 2 unexpectedly empty: %#v", got)
+	}
+}
+
+func TestWrapToastBodyEmptyInput(t *testing.T) {
+	got := wrapToastBody("", 20)
+	for i, l := range got {
+		if l != "" {
+			t.Fatalf("line %d = %q, want \"\"", i, l)
+		}
+	}
+}

internal/app/token.go

@@ -0,0 +1,63 @@
+package app
+
+import (
+	"crypto/rand"
+	"encoding/base64"
+	"fmt"
+	"os"
+	"path/filepath"
+	"strings"
+)
+
+func ClientTokenPath() (string, error) {
+	base := os.Getenv("XDG_DATA_HOME")
+	if base == "" {
+		home, err := os.UserHomeDir()
+		if err != nil {
+			return "", err
+		}
+		base = filepath.Join(home, ".local", "share")
+	}
+	return filepath.Join(base, "patterm", "clients", "token"), nil
+}
+
+func LoadClientToken() (string, error) {
+	path, err := ClientTokenPath()
+	if err != nil {
+		return "", err
+	}
+	b, err := os.ReadFile(path)
+	if err != nil {
+		return "", err
+	}
+	return strings.TrimSpace(string(b)), nil
+}
+
+func LoadOrCreateClientToken() (string, error) {
+	if token, err := LoadClientToken(); err == nil && token != "" {
+		return token, nil
+	}
+	token, err := generateClientToken()
+	if err != nil {
+		return "", err
+	}
+	path, err := ClientTokenPath()
+	if err != nil {
+		return "", err
+	}
+	if err := os.MkdirAll(filepath.Dir(path), 0o700); err != nil {
+		return "", err
+	}
+	if err := os.WriteFile(path, []byte(token+"\n"), 0o600); err != nil {
+		return "", err
+	}
+	return token, nil
+}
+
+func generateClientToken() (string, error) {
+	var b [32]byte
+	if _, err := rand.Read(b[:]); err != nil {
+		return "", fmt.Errorf("token: random: %w", err)
+	}
+	return base64.RawURLEncoding.EncodeToString(b[:]), nil
+}

internal/app/tree.go

@@ -1,5 +1,19 @@
 package app
 
+import "github.com/hjbdev/patterm/internal/scratchpad"
+
+// navEntry is one row in the unified sidebar navigation list. Exactly
+// one of childID or pad is set. childID points at a Child by ID; pad
+// names a scratchpad entry. Empty zero-value means "no target".
+type navEntry struct {
+	childID string
+	pad     string
+}
+
+func (n navEntry) empty() bool   { return n.childID == "" && n.pad == "" }
+func (n navEntry) isPad() bool   { return n.pad != "" }
+func (n navEntry) isChild() bool { return n.childID != "" }
+
 // visibleAgentTree returns the running entries under the active agent
 // tab (root agent + its sub-agents). With the new Processes pane,
 // command processes live in their own section and never show up here —
@@ -82,17 +96,24 @@ func firstRunningAgentID(children []*Child) string {
 }
 
 // processList returns every top-level command/terminal entry in spawn
-// order, regardless of running state. The Processes sidebar section
-// keeps showing exited entries so the user can see what just died (and
-// because Session retains KindCommand entries for restart).
+// order. Exited KindCommand entries remain visible so the user can see
+// what just died and reach restart_process; exited KindTerminal entries
+// are filtered out because terminals are ephemeral and have no restart
+// path (Session also drops them in reapChild — this filter is defensive
+// for any window between exit and deletion).
 func processList(children []*Child) []*Child {
 	out := make([]*Child, 0, len(children))
 	for _, c := range children {
 		if c.ParentID != "" {
 			continue
 		}
-		if c.Kind == KindCommand || c.Kind == KindTerminal {
+		switch c.Kind {
+		case KindCommand:
 			out = append(out, c)
+		case KindTerminal:
+			if c.Status() == StatusRunning {
+				out = append(out, c)
+			}
 		}
 	}
 	return out
@@ -192,9 +213,6 @@ func currentTabFlat(children []*Child, focusID string) []*Child {
 func sidebarNavList(children []*Child, activeAgentID string) []*Child {
 	out := make([]*Child, 0, 8)
 	for _, c := range processList(children) {
-		if c.Status() != StatusRunning {
-			continue
-		}
 		out = append(out, c)
 	}
 	for _, c := range visibleAgentTree(children, activeAgentID) {
@@ -203,14 +221,77 @@ func sidebarNavList(children []*Child, activeAgentID string) []*Child {
 	return out
 }
 
-// nextChildID returns the id `step` positions away from the current
-// focus in the combined Processes + active-agent-tree navigation list,
-// wrapping at both ends. Empty when there's nothing else to land on.
+// sidebarNav returns the combined Processes + Agent Tree + Scratchpads
+// navigation list. Scratchpads always appear after children so the
+// existing "step past the tree" expectation still holds.
+func sidebarNav(children []*Child, activeAgentID string, pads []scratchpad.Entry) []navEntry {
+	flat := sidebarNavList(children, activeAgentID)
+	out := make([]navEntry, 0, len(flat)+len(pads))
+	for _, c := range flat {
+		out = append(out, navEntry{childID: c.ID})
+	}
+	for _, p := range pads {
+		out = append(out, navEntry{pad: p.Name})
+	}
+	return out
+}
+
+// nextNavEntry returns the entry `step` positions away from the
+// current focus in the unified nav list. Either focusChildID or
+// focusPad will be set (or both empty for "nothing focused yet").
+// Empty when there's nothing else to land on.
+func nextNavEntry(children []*Child, focusChildID, focusPad, activeAgentID string, pads []scratchpad.Entry, step int) navEntry {
+	flat := sidebarNav(children, activeAgentID, pads)
+	if len(flat) == 0 {
+		return navEntry{}
+	}
+	matches := func(e navEntry) bool {
+		if focusPad != "" && e.pad != "" {
+			return e.pad == focusPad
+		}
+		if focusChildID != "" && e.childID != "" {
+			return e.childID == focusChildID
+		}
+		return false
+	}
+	if len(flat) == 1 {
+		if matches(flat[0]) {
+			return navEntry{}
+		}
+		return flat[0]
+	}
+	idx := -1
+	for i, e := range flat {
+		if matches(e) {
+			idx = i
+			break
+		}
+	}
+	if idx < 0 {
+		idx = 0
+	}
+	idx = (idx + step) % len(flat)
+	if idx < 0 {
+		idx += len(flat)
+	}
+	if matches(flat[idx]) {
+		return navEntry{}
+	}
+	return flat[idx]
+}
+
+// nextChildID is retained for tests; it ignores scratchpads.
 func nextChildID(children []*Child, focusID, activeAgentID string, step int) string {
 	flat := sidebarNavList(children, activeAgentID)
-	if len(flat) < 2 {
+	if len(flat) == 0 {
 		return ""
 	}
+	if len(flat) == 1 {
+		if flat[0].ID == focusID {
+			return ""
+		}
+		return flat[0].ID
+	}
 	idx := -1
 	for i, c := range flat {
 		if c.ID == focusID {

internal/app/tree_test.go

@@ -125,6 +125,15 @@ func TestSidebarNavListIncludesProcessesAboveAgentTree(t *testing.T) {
 	}
 }
 
+func TestSidebarNavListIncludesExitedProcesses(t *testing.T) {
+	p := testProcess("p1", "shell", StatusExited)
+	r := testAgent("a1", "claude", "", StatusRunning)
+	flat := sidebarNavList([]*Child{p, r}, "a1")
+	if len(flat) != 2 || flat[0].ID != "p1" || flat[1].ID != "a1" {
+		t.Fatalf("flat = %v, want exited process then active agent", childIDs(flat))
+	}
+}
+
 func TestNextChildIDWalksProcessesThenAgentTree(t *testing.T) {
 	p1 := testProcess("p1", "bun", StatusRunning)
 	r := testAgent("a1", "claude", "", StatusRunning)
@@ -140,6 +149,13 @@ func TestNextChildIDWalksProcessesThenAgentTree(t *testing.T) {
 	}
 }
 
+func TestNextChildIDCanEnterSingleExitedProcessFromNoFocus(t *testing.T) {
+	p := testProcess("p1", "shell", StatusExited)
+	if got := nextChildID([]*Child{p}, "", "", +1); got != "p1" {
+		t.Fatalf("empty focus -> exited process: %q want p1", got)
+	}
+}
+
 func TestVisibleAgentTreeExcludesTopLevelCommands(t *testing.T) {
 	p := testProcess("p1", "bun", StatusRunning)
 	r := testAgent("a1", "claude", "", StatusRunning)

internal/app/viewport_renderer.go

@@ -10,11 +10,15 @@ import (
 // viewportRenderer rewrites child PTY output so it lands inside the
 // main viewport instead of controlling patterm's full host terminal.
 type viewportRenderer struct {
-	mu      sync.Mutex
-	shifter *cursorShifter
-	layout  terminalLayout
-	row     int
-	col     int
+	mu           sync.Mutex
+	shifter      *cursorShifter
+	layout       terminalLayout
+	row          int
+	col          int
+	scrollTop    int
+	scrollBottom int
+	originMode   bool
+	lrMarginMode bool
 
 	state   viewportState
 	buf     []byte
@@ -22,12 +26,21 @@ type viewportRenderer struct {
 
 	// scrolled is set when the chunk contained an escape that shifts
 	// content row-wise within the host's scroll region — RI / IND /
-	// NEL / SU / SD / IL / DL. DECSTBM constrains rows but not columns,
-	// so these scrolls drag the right-hand sidebar content with them.
+	// NEL / SU / SD / IL / DL, or LF / VT / FF at the bottom margin.
+	// DECSTBM constrains rows but not columns, so these scrolls drag the
+	// right-hand sidebar content with them.
 	// OnPTYOut consumes the flag and invalidates the sidebar chrome
 	// cache so the next drawSidebar repaints over the clobber.
 	scrolled bool
 
+	// childOnAlt tracks whether the focused child has entered its
+	// alternate screen (via ?47 / ?1047 / ?1049). Used to gate mouse-
+	// tracking-mode forwarding to the host: filter on primary so
+	// patterm's wheel-scrollback stays armed, forward on alt so codex
+	// (which disables mouse) lets the user select text and vim (which
+	// enables it) still gets mouse events.
+	childOnAlt bool
+
 	// skipUTF8 is set when the current multi-byte UTF-8 character started
 	// past the viewport's right edge. The starter byte was dropped, so
 	// the remaining continuation bytes must be dropped too instead of
@@ -50,12 +63,24 @@ const (
 )
 
 func newViewportRenderer(l terminalLayout) *viewportRenderer {
-	return &viewportRenderer{
+	vr := &viewportRenderer{
 		shifter: newCursorShifter(int(l.mainTop)-1, int(l.childRows()), int(l.childCols())),
 		layout:  l,
 		row:     1,
 		col:     1,
 	}
+	vr.resetScrollRegion()
+	return vr
+}
+
+// SetChildOnAlt seeds the renderer's view of the focused child's screen
+// side. Used when a new renderer is constructed for an already-running
+// child whose alt-screen transition we missed, so subsequent mouse-mode
+// toggles are filtered/forwarded according to the right side.
+func (vr *viewportRenderer) SetChildOnAlt(onAlt bool) {
+	vr.mu.Lock()
+	defer vr.mu.Unlock()
+	vr.childOnAlt = onAlt
 }
 
 func (vr *viewportRenderer) SetLayout(l terminalLayout) {
@@ -63,14 +88,47 @@ func (vr *viewportRenderer) SetLayout(l terminalLayout) {
 	defer vr.mu.Unlock()
 	vr.layout = l
 	vr.shifter.SetGeometry(int(l.mainTop)-1, int(l.childRows()), int(l.childCols()))
+	vr.resetScrollRegion()
 }
 
 func (vr *viewportRenderer) Render(in []byte) []byte {
 	vr.mu.Lock()
 	defer vr.mu.Unlock()
 	vr.pending.Reset()
-	for _, b := range in {
-		vr.feed(b)
+	// Fast path: while we're in vpNormal and have a run of plain ASCII
+	// printables that fit the remaining column budget, copy en bloc
+	// instead of round-tripping each byte through the feed state
+	// machine. UTF-8 leaders and any control byte fall back to the
+	// per-byte path so the cursor/skipUTF8/clamp logic stays exact.
+	for i := 0; i < len(in); {
+		if vr.state == vpNormal {
+			maxCol := int(vr.layout.childCols())
+			if maxCol > 0 && vr.col >= 1 && vr.col <= maxCol {
+				budget := maxCol - vr.col + 1
+				j := i
+				for j < len(in) && budget > 0 {
+					b := in[j]
+					// Pure ASCII printables only — any control byte
+					// (0x1b ESC included), UTF-8 leader, or trailer
+					// kicks back to the state machine.
+					if b < 0x20 || b == 0x7f || b >= 0x80 {
+						break
+					}
+					j++
+					budget--
+				}
+				if j-i >= 4 {
+					vr.pending.Write(in[i:j])
+					vr.col += j - i
+					vr.skipUTF8 = false
+					vr.clampCursor()
+					i = j
+					continue
+				}
+			}
+		}
+		vr.feed(in[i])
+		i++
 	}
 	return []byte(vr.pending.String())
 }
@@ -82,11 +140,10 @@ func (vr *viewportRenderer) ClearViewport() []byte {
 }
 
 // TookScrollAction reports whether the most recent Render emitted (or
-// forwarded) a scroll-triggering escape — RI / IND / NEL / SU / SD /
-// IL / DL — since the previous call. The flag is reset on read.
-// Callers use it to invalidate sidebar-cache state, because the host's
-// scroll region spans the full row width and any scroll there drags
-// the sidebar content downward.
+// forwarded) a scroll action since the previous call. Callers use it
+// to invalidate sidebar-cache state, because the host's scroll region
+// spans the full row width and any scroll there drags the sidebar
+// content vertically.
 func (vr *viewportRenderer) TookScrollAction() bool {
 	vr.mu.Lock()
 	defer vr.mu.Unlock()
@@ -187,12 +244,74 @@ func (vr *viewportRenderer) emitCSI() {
 	params := vr.buf[2 : len(vr.buf)-1]
 
 	if final == 'h' || final == 'l' {
+		if isOriginMode(params) {
+			vr.setOriginMode(final == 'h')
+			vr.emitCursorPosition(vr.row, vr.col)
+			return
+		}
+		if isLeftRightMarginMode(params) {
+			vr.lrMarginMode = final == 'h'
+			return
+		}
 		if isAltScreenMode(params) {
+			// Track the child's screen side so we know whether to filter
+			// or forward subsequent mouse-mode toggles. Entering alt
+			// disables host mouse reporting by default so codex (and
+			// any other alt-screen TUI that doesn't request mouse)
+			// allows the user to click-drag to select text. Alt-screen
+			// TUIs that want mouse (vim, less with -X) re-enable it
+			// via ?1000h after switching to alt — the forwarder below
+			// passes that through. Leaving alt re-arms host mouse for
+			// primary-screen wheel-scrollback.
+			wasAlt := vr.childOnAlt
+			vr.childOnAlt = final == 'h'
+			if !wasAlt && vr.childOnAlt {
+				vr.pending.WriteString("\x1b[?1000l\x1b[?1006l")
+			}
+			if wasAlt && !vr.childOnAlt {
+				vr.pending.WriteString("\x1b[?1000h\x1b[?1006h")
+			}
+			return
+		}
+		if isMouseTrackingMode(params) {
+			// On the child's primary screen patterm owns mouse reporting so
+			// wheel events keep flowing for in-pane scrollback — drop the
+			// child's toggle. On the alt screen the child should be free
+			// to enable mouse (vim, less) or disable it (codex); we forward
+			// the toggle to the host so click-and-drag selection works for
+			// alt-screen TUIs that don't want mouse, and mouse-aware ones
+			// still see the events they need.
+			if vr.childOnAlt {
+				vr.pending.Write(vr.buf)
+			}
 			return
 		}
 	}
 
+	if final == 's' && vr.lrMarginMode {
+		return
+	}
+
 	switch final {
+	case 'H', 'f':
+		r, c, ok := parseTwoParams(params)
+		if !ok {
+			vr.pending.Write(vr.shifter.Shift(vr.buf))
+			return
+		}
+		vr.row = vr.originRow(r)
+		vr.col = c
+		vr.emitCursorPosition(vr.row, c)
+		vr.clampCursor()
+	case 'd':
+		r, ok := parseOneParam(params, 1)
+		if !ok {
+			vr.pending.Write(vr.shifter.Shift(vr.buf))
+			return
+		}
+		vr.row = vr.originRow(r)
+		vr.pending.Write(vr.shifter.Shift([]byte(fmt.Sprintf("\x1b[%dd", vr.row))))
+		vr.clampCursor()
 	case 'J':
 		n, ok := parseOneParam(params, 0)
 		if !ok {
@@ -225,10 +344,85 @@ func (vr *viewportRenderer) emitCSI() {
 		// the sidebar is repainted afterwards.
 		vr.pending.Write(vr.shifter.Shift(vr.buf))
 		vr.scrolled = true
+	case 'r':
+		vr.pending.Write(vr.shifter.Shift(vr.buf))
+		if vr.trackScrollRegion(params) {
+			vr.emitHomeAfterScrollRegion()
+		}
+	case 'A', 'B', 'E', 'F':
+		// Relative cursor moves: CUU (A) / CUD (B) / CNL (E) / CPL (F).
+		// The cursor shifter only rewrites absolute positioning, so a
+		// child that asks the cursor to "go up 50" from viewport row 1
+		// would walk the host cursor into the tab bar (and the next
+		// printable would write there). Clamp the step using the
+		// renderer's tracked row so the host cursor stays inside the
+		// viewport. E / F additionally home the column to 1.
+		vr.emitRelativeRowMove(final, params)
+		return
 	default:
 		vr.pending.Write(vr.shifter.Shift(vr.buf))
 	}
-	vr.trackCSI(final, params)
+	if final != 'H' && final != 'f' && final != 'd' && final != 'r' {
+		vr.trackCSI(final, params)
+	}
+}
+
+// emitRelativeRowMove rewrites CSI A / B / E / F so the resulting host
+// cursor stays within rows 1..childRows in viewport coordinates. The
+// renderer already tracks vr.row for clear-line bookkeeping; reusing
+// that here avoids a second cursor model. n is normalized — a step of
+// 0 is treated as 1 to match xterm. After clamping, if the effective
+// step is zero we drop the sequence (the cursor is already pinned to
+// the boundary). E / F also move the cursor to column 1 even when no
+// row step is emitted.
+func (vr *viewportRenderer) emitRelativeRowMove(final byte, params []byte) {
+	n, ok := parseOneParam(params, 1)
+	if !ok {
+		vr.pending.Write(vr.shifter.Shift(vr.buf))
+		return
+	}
+	if n <= 0 {
+		n = 1
+	}
+	rows := int(vr.layout.childRows())
+	if rows < 1 {
+		rows = 1
+	}
+	row := vr.row
+	if row < 1 {
+		row = 1
+	}
+	if row > rows {
+		row = rows
+	}
+	up := final == 'A' || final == 'F'
+	var safe int
+	if up {
+		safe = row - 1
+	} else {
+		safe = rows - row
+	}
+	if safe < 0 {
+		safe = 0
+	}
+	if n > safe {
+		n = safe
+	}
+	if n > 0 {
+		if up {
+			vr.row -= n
+		} else {
+			vr.row += n
+		}
+		fmt.Fprintf(&vr.pending, "\x1b[%d%c", n, final)
+	}
+	if final == 'E' || final == 'F' {
+		// CNL / CPL anchor the column at 1 regardless of whether the
+		// row step was clamped to zero, matching xterm.
+		vr.col = 1
+		vr.pending.WriteByte('\r')
+	}
+	vr.clampCursor()
 }
 
 func isAltScreenMode(params []byte) bool {
@@ -245,6 +439,52 @@ func isAltScreenMode(params []byte) bool {
 	return false
 }
 
+func isOriginMode(params []byte) bool {
+	s := string(params)
+	if !strings.HasPrefix(s, "?") {
+		return false
+	}
+	for _, p := range strings.Split(strings.TrimPrefix(s, "?"), ";") {
+		if p == "6" {
+			return true
+		}
+	}
+	return false
+}
+
+func isLeftRightMarginMode(params []byte) bool {
+	s := string(params)
+	if !strings.HasPrefix(s, "?") {
+		return false
+	}
+	for _, p := range strings.Split(strings.TrimPrefix(s, "?"), ";") {
+		if p == "69" {
+			return true
+		}
+	}
+	return false
+}
+
+// isMouseTrackingMode reports whether any of the modes in a CSI ? … h/l
+// is a mouse-tracking or mouse-encoding DEC private mode. The host runs
+// with SGR mouse reporting permanently armed; we drop the child's set/
+// reset for these modes from the host stream so wheel events keep
+// reaching patterm.
+func isMouseTrackingMode(params []byte) bool {
+	s := string(params)
+	if !strings.HasPrefix(s, "?") {
+		return false
+	}
+	for _, p := range strings.Split(strings.TrimPrefix(s, "?"), ";") {
+		switch p {
+		case "9", "1000", "1001", "1002", "1003", "1004",
+			"1005", "1006", "1007", "1015", "1016":
+			return true
+		}
+	}
+	return false
+}
+
 func (vr *viewportRenderer) clearViewport() string {
 	var b strings.Builder
 	b.WriteString("\x1b7")
@@ -326,6 +566,69 @@ func (vr *viewportRenderer) clearLine(n int) string {
 	}
 }
 
+func (vr *viewportRenderer) resetScrollRegion() {
+	vr.scrollTop = 1
+	vr.scrollBottom = int(vr.layout.childRows())
+	if vr.scrollBottom < 1 {
+		vr.scrollBottom = 1
+	}
+}
+
+func (vr *viewportRenderer) setOriginMode(on bool) {
+	vr.originMode = on
+	if on {
+		vr.row = vr.scrollTop
+	} else {
+		vr.row = 1
+	}
+	vr.col = 1
+	vr.clampCursor()
+}
+
+func (vr *viewportRenderer) originRow(row int) int {
+	if row < 1 {
+		row = 1
+	}
+	if !vr.originMode {
+		return row
+	}
+	row = vr.scrollTop + row - 1
+	if row < vr.scrollTop {
+		row = vr.scrollTop
+	}
+	if row > vr.scrollBottom {
+		row = vr.scrollBottom
+	}
+	return row
+}
+
+func (vr *viewportRenderer) homeAfterScrollRegion() {
+	if vr.originMode {
+		vr.row = vr.scrollTop
+	} else {
+		vr.row = 1
+	}
+	vr.col = 1
+	vr.clampCursor()
+}
+
+func (vr *viewportRenderer) emitHomeAfterScrollRegion() {
+	vr.homeAfterScrollRegion()
+	vr.emitCursorPosition(vr.row, vr.col)
+}
+
+func (vr *viewportRenderer) emitCursorPosition(row, col int) {
+	vr.pending.Write(vr.shifter.Shift([]byte(fmt.Sprintf("\x1b[%d;%dH", row, col))))
+}
+
+func (vr *viewportRenderer) lineFeed() {
+	if vr.row >= vr.scrollTop && vr.row == vr.scrollBottom {
+		vr.scrolled = true
+		return
+	}
+	vr.row++
+}
+
 // feedPrintable handles one non-ESC byte in the vpNormal state. It both
 // advances vr's cursor model and decides whether the byte should be
 // forwarded to the host. Bytes that would land past the viewport's
@@ -342,8 +645,8 @@ func (vr *viewportRenderer) feedPrintable(b byte) {
 		switch b {
 		case '\r':
 			vr.col = 1
-		case '\n':
-			vr.row++
+		case '\n', '\v', '\f':
+			vr.lineFeed()
 		case '\b':
 			if vr.col > 1 {
 				vr.col--
@@ -405,7 +708,7 @@ func (vr *viewportRenderer) trackCSI(final byte, params []byte) {
 	case 'H', 'f':
 		r, c, ok := parseTwoParams(params)
 		if ok {
-			vr.row, vr.col = r, c
+			vr.row, vr.col = vr.originRow(r), c
 		}
 	case 'G', '`':
 		c, ok := parseOneParam(params, 1)
@@ -415,7 +718,7 @@ func (vr *viewportRenderer) trackCSI(final byte, params []byte) {
 	case 'd':
 		r, ok := parseOneParam(params, 1)
 		if ok {
-			vr.row = r
+			vr.row = vr.originRow(r)
 		}
 	case 'A':
 		n, ok := parseOneParam(params, 1)
@@ -437,10 +740,41 @@ func (vr *viewportRenderer) trackCSI(final byte, params []byte) {
 		if ok {
 			vr.col -= n
 		}
+	case 'r':
+		if vr.trackScrollRegion(params) {
+			vr.homeAfterScrollRegion()
+		}
 	}
 	vr.clampCursor()
 }
 
+func (vr *viewportRenderer) trackScrollRegion(params []byte) bool {
+	if len(params) == 0 {
+		vr.resetScrollRegion()
+		return true
+	}
+	top, bottom, ok := parseTwoParams(params)
+	if !ok {
+		return false
+	}
+	maxRows := int(vr.layout.childRows())
+	if maxRows < 1 {
+		maxRows = 1
+	}
+	if top < 1 {
+		top = 1
+	}
+	if bottom < 1 || bottom > maxRows {
+		bottom = maxRows
+	}
+	if top >= bottom {
+		return false
+	}
+	vr.scrollTop = top
+	vr.scrollBottom = bottom
+	return true
+}
+
 func (vr *viewportRenderer) clampCursor() {
 	if vr.row < 1 {
 		vr.row = 1

internal/app/viewport_renderer_test.go

@@ -16,7 +16,7 @@ func bytesRepeat(b byte, n int) []byte {
 func TestViewportRendererShiftsCursor(t *testing.T) {
 	vr := newViewportRenderer(newTerminalLayout(120, 40))
 	got := string(vr.Render([]byte("\x1b[H")))
-	if got != "\x1b[3;1H" {
+	if got != "\x1b[4;1H" {
 		t.Fatalf("CUP home: got %q", got)
 	}
 }
@@ -24,23 +24,87 @@ func TestViewportRendererShiftsCursor(t *testing.T) {
 func TestViewportRendererSwallowsAltScreenToggles(t *testing.T) {
 	vr := newViewportRenderer(newTerminalLayout(120, 40))
 	got := string(vr.Render([]byte("a\x1b[?1049hb\x1b[?1049lc")))
+	// The ?1049h/l toggles themselves must not reach the host (patterm
+	// owns its own alt screen). On the transition we re-sync host mouse
+	// reporting so codex (which doesn't request mouse) lets the user
+	// drag-select; leaving alt re-arms it for primary-screen wheel
+	// scrollback.
+	want := "a\x1b[?1000l\x1b[?1006lb\x1b[?1000h\x1b[?1006hc"
+	if got != want {
+		t.Fatalf("alt-screen toggles: got %q want %q", got, want)
+	}
+}
+
+func TestViewportRendererMouseTrackingFilteredOnPrimary(t *testing.T) {
+	vr := newViewportRenderer(newTerminalLayout(120, 40))
+	got := string(vr.Render([]byte("a\x1b[?1000lb\x1b[?1000hc")))
 	if got != "abc" {
-		t.Fatalf("alt-screen toggles: got %q", got)
+		t.Fatalf("mouse mode on primary should be filtered: got %q", got)
+	}
+}
+
+func TestViewportRendererMouseTrackingForwardedOnAlt(t *testing.T) {
+	vr := newViewportRenderer(newTerminalLayout(120, 40))
+	// Enter alt; subsequent mouse-mode toggles should reach the host so
+	// alt-screen TUIs (vim, less) can run with mouse on, and selection-
+	// using ones (codex) stay with mouse off.
+	got := string(vr.Render([]byte("\x1b[?1049h\x1b[?1000lx\x1b[?1000hy")))
+	if !strings.Contains(got, "\x1b[?1000l") {
+		t.Fatalf("alt-screen mouse disable should reach host: %q", got)
+	}
+	if !strings.Contains(got, "\x1b[?1000h") {
+		t.Fatalf("alt-screen mouse enable should reach host: %q", got)
+	}
+}
+
+func TestViewportRendererSwallowsOriginModeToggles(t *testing.T) {
+	vr := newViewportRenderer(newTerminalLayout(120, 40))
+	got := string(vr.Render([]byte("a\x1b[?6hb\x1b[?6lc")))
+	if strings.Contains(got, "\x1b[?6h") || strings.Contains(got, "\x1b[?6l") {
+		t.Fatalf("origin-mode toggles leaked to host: %q", got)
+	}
+	if !strings.Contains(got, "a") || !strings.Contains(got, "b") || !strings.Contains(got, "c") {
+		t.Fatalf("origin-mode toggles should not drop surrounding text: got %q", got)
+	}
+	if strings.Count(got, "\x1b[4;1H") != 2 {
+		t.Fatalf("origin-mode set/reset should home inside the viewport twice: got %q", got)
+	}
+}
+
+func TestViewportRendererSwallowsLeftRightMarginMode(t *testing.T) {
+	vr := newViewportRenderer(newTerminalLayout(120, 40))
+	got := string(vr.Render([]byte("a\x1b[?69h\x1b[10;80sb\x1b[?69lc")))
+	if strings.Contains(got, "\x1b[?69h") || strings.Contains(got, "\x1b[10;80s") || strings.Contains(got, "\x1b[?69l") {
+		t.Fatalf("left/right margin controls leaked to host: %q", got)
+	}
+	if got != "abc" {
+		t.Fatalf("left/right margin controls should be swallowed without dropping text: got %q", got)
+	}
+}
+
+func TestViewportRendererOriginModeCUPUsesScrollTop(t *testing.T) {
+	vr := newViewportRenderer(newTerminalLayout(120, 40))
+	got := string(vr.Render([]byte("\x1b[5;10r\x1b[?6h\x1b[1;1H")))
+	if strings.Contains(got, "\x1b[?6h") {
+		t.Fatalf("origin-mode set leaked to host: %q", got)
+	}
+	if !strings.Contains(got, "\x1b[8;1H") {
+		t.Fatalf("CUP row 1 in origin mode should land at scrollTop row 5 shifted to host row 8: got %q", got)
 	}
 }
 
 func TestViewportRendererClearScreenIsViewportOnly(t *testing.T) {
-	// hostRows=7 leaves four viewport rows after the 2-row tab bar and
+	// hostRows=7 leaves three viewport rows after the 3-row tab bar and
 	// 1-row status reservation.
 	vr := newViewportRenderer(newTerminalLayout(20, 7))
 	got := string(vr.Render([]byte("\x1b[2J")))
 	if strings.Contains(got, "\x1b[2J") {
 		t.Fatalf("host clear-screen leaked through: %q", got)
 	}
-	if strings.Count(got, "\x1b[20X") != 4 {
+	if strings.Count(got, "\x1b[20X") != 3 {
 		t.Fatalf("clear rows: got %q", got)
 	}
-	if !strings.Contains(got, "\x1b[3;1H") || !strings.Contains(got, "\x1b[6;1H") {
+	if !strings.Contains(got, "\x1b[4;1H") || !strings.Contains(got, "\x1b[6;1H") {
 		t.Fatalf("clear did not target viewport rows: %q", got)
 	}
 }
@@ -76,13 +140,12 @@ func TestViewportRendererClearToEndIsViewportOnly(t *testing.T) {
 		t.Fatalf("host clear-to-end leaked through: %q", got)
 	}
 	// childCols == 19 (40 cols - 28 sidebar - 1 gap - 0-index fudge).
-	// Each of the 4 viewport rows should get a 19-cell erase.
 	// childCols == 11 with hostCols=40 (28 sidebar + 1 gap reserved).
-	// 4 viewport rows, but the cursor row uses ECH at cursor (col 1),
-	// so we expect 4 erases of 11 cells each.
+	// 3 viewport rows, but the cursor row uses ECH at cursor (col 1),
+	// so we expect 3 erases of 11 cells each.
 	count := strings.Count(got, "\x1b[11X")
-	if count != 4 {
-		t.Fatalf("expected 4 ECH-11 sequences, got %d in %q", count, got)
+	if count != 3 {
+		t.Fatalf("expected 3 ECH-11 sequences, got %d in %q", count, got)
 	}
 }
 
@@ -118,7 +181,7 @@ func TestViewportRendererClampsCUPColumn(t *testing.T) {
 	// column so the host cursor never lands in the sidebar.
 	vr := newViewportRenderer(newTerminalLayout(120, 40))
 	got := string(vr.Render([]byte("\x1b[5;95H")))
-	if !strings.Contains(got, "\x1b[7;91H") {
+	if !strings.Contains(got, "\x1b[8;91H") {
 		t.Fatalf("CUP col 95 should clamp to 91 (childCols): got %q", got)
 	}
 }
@@ -211,6 +274,101 @@ func TestViewportRendererFlagsScrollVerbs(t *testing.T) {
 	}
 }
 
+func TestViewportRendererFlagsLineFeedAtViewportBottomAsScrolling(t *testing.T) {
+	vr := newViewportRenderer(newTerminalLayout(120, 40))
+	_ = vr.Render([]byte("\x1b[36;1H\n"))
+	if !vr.TookScrollAction() {
+		t.Fatalf("LF at viewport bottom should flag scroll")
+	}
+}
+
+func TestViewportRendererDoesNotFlagLineFeedBeforeViewportBottom(t *testing.T) {
+	vr := newViewportRenderer(newTerminalLayout(120, 40))
+	_ = vr.Render([]byte("\x1b[35;1H\n"))
+	if vr.TookScrollAction() {
+		t.Fatalf("LF before viewport bottom should not flag scroll")
+	}
+}
+
+func TestViewportRendererFlagsLineFeedAtCustomScrollBottom(t *testing.T) {
+	vr := newViewportRenderer(newTerminalLayout(120, 40))
+	_ = vr.Render([]byte("\x1b[5;10r\x1b[9;1H\n"))
+	if vr.TookScrollAction() {
+		t.Fatalf("LF before custom scroll bottom should not flag scroll")
+	}
+	_ = vr.Render([]byte("\n"))
+	if !vr.TookScrollAction() {
+		t.Fatalf("LF at custom scroll bottom should flag scroll")
+	}
+}
+
+// Long claude sessions can leave the child cursor at viewport row 1 and
+// then emit CSI A (cursor up) with a large step before redrawing. The
+// raw CSI A would walk the host cursor into the tab bar; the next
+// printable would then write into row 1 / row 2. Clamp the step at the
+// viewport top so the host cursor stays inside the viewport.
+func TestViewportRendererClampsCUUAtViewportTop(t *testing.T) {
+	vr := newViewportRenderer(newTerminalLayout(120, 40))
+	// CUP to viewport row 1 then CUU by 50.
+	got := string(vr.Render([]byte("\x1b[1;1H\x1b[50ACLOBBER")))
+	if !strings.Contains(got, "\x1b[4;1H") {
+		t.Fatalf("expected CUP shifted to mainTop: got %q", got)
+	}
+	// The CUU should have been swallowed (n clamped to 0 from row 1).
+	if strings.Contains(got, "\x1b[50A") {
+		t.Fatalf("CUU 50 from viewport row 1 leaked: got %q", got)
+	}
+	// And the subsequent printables should land inside the viewport,
+	// not above it.
+	if !strings.Contains(got, "CLOBBER") {
+		t.Fatalf("printables should still be emitted after clamped CUU: got %q", got)
+	}
+}
+
+func TestViewportRendererClampsCUUPartial(t *testing.T) {
+	vr := newViewportRenderer(newTerminalLayout(120, 40))
+	// CUP to viewport row 5, then CUU by 50 → safe step is 4.
+	got := string(vr.Render([]byte("\x1b[5;1H\x1b[50A")))
+	if !strings.Contains(got, "\x1b[4A") {
+		t.Fatalf("CUU 50 from row 5 should clamp to 4: got %q", got)
+	}
+	if strings.Contains(got, "\x1b[50A") {
+		t.Fatalf("unclamped CUU leaked: got %q", got)
+	}
+}
+
+func TestViewportRendererClampsCUDAtViewportBottom(t *testing.T) {
+	// childRows=36 for layout(120, 40). Park cursor at row 36, ask for
+	// 10 down → safe step is 0.
+	vr := newViewportRenderer(newTerminalLayout(120, 40))
+	got := string(vr.Render([]byte("\x1b[36;1H\x1b[10B")))
+	if strings.Contains(got, "\x1b[10B") {
+		t.Fatalf("CUD past viewport bottom should be dropped: got %q", got)
+	}
+}
+
+func TestViewportRendererClampsCPLAndHomesColumn(t *testing.T) {
+	vr := newViewportRenderer(newTerminalLayout(120, 40))
+	// CUP to row 1 col 50 then CPL by 5 → step clamped to 0, but col
+	// must still reset to 1 (CR emitted).
+	got := string(vr.Render([]byte("\x1b[1;50H\x1b[5F")))
+	if strings.Contains(got, "\x1b[5F") {
+		t.Fatalf("CPL 5 from row 1 should not leak: got %q", got)
+	}
+	if !strings.Contains(got, "\r") {
+		t.Fatalf("CPL should home column to 1 with CR: got %q", got)
+	}
+}
+
+func TestViewportRendererClampsCNL(t *testing.T) {
+	vr := newViewportRenderer(newTerminalLayout(120, 40))
+	// CUP to row 34 then CNL by 50 → safe step is 2 (childRows-34).
+	got := string(vr.Render([]byte("\x1b[34;10H\x1b[50E")))
+	if !strings.Contains(got, "\x1b[2E") {
+		t.Fatalf("CNL 50 from row 34 should clamp to 2: got %q", got)
+	}
+}
+
 func TestViewportRendererForwardsRIVerbatim(t *testing.T) {
 	// We rely on the host terminal performing the scroll inside the
 	// DECSTBM region; the renderer must not eat or transform RI. If a

internal/harness/input.go

@@ -30,10 +30,29 @@ func EncodeChord(name string) ([]byte, error) {
 		return []byte{0x10}, nil
 	case "ctrl-u":
 		return []byte{0x15}, nil
+	case "ctrl-a":
+		return []byte{0x01}, nil
+	case "ctrl-d":
+		return []byte{0x04}, nil
+	case "ctrl-s":
+		return []byte{0x13}, nil
+	case "ctrl-w":
+		return []byte{0x17}, nil
+	case "ctrl-r":
+		return []byte{0x12}, nil
+	case "ctrl-b":
+		return []byte{0x02}, nil
 	case "tab":
 		return []byte{'\t'}, nil
 	case "space":
 		return []byte{' '}, nil
+	case "wheel-up":
+		// SGR-encoded scroll-wheel up at row/col 1,1. patterm enables
+		// 1006 mouse mode while a scratchpad is focused, so this is the
+		// form the host terminal would deliver.
+		return []byte("\x1b[<64;1;1M"), nil
+	case "wheel-down":
+		return []byte("\x1b[<65;1;1M"), nil
 	}
 	return nil, fmt.Errorf("unknown chord %q", name)
 }

internal/harness/restart_persist_test.go

@@ -0,0 +1,187 @@
+package harness
+
+import (
+	"encoding/json"
+	"fmt"
+	"os"
+	"path/filepath"
+	"testing"
+	"time"
+
+	pkgpty "github.com/hjbdev/patterm/internal/pty"
+	"github.com/hjbdev/patterm/internal/vt"
+)
+
+// TestRestartRestoresUserCommandProcess verifies that a process the
+// user spawned in one patterm run reappears after the binary is
+// restarted against the same XDG dirs / project dir. SPEC §2 keeps
+// runs ephemeral except for the persisted-process state file:
+// processes.json under $XDG_DATA_HOME/patterm/projects/<key>/.
+func TestRestartRestoresUserCommandProcess(t *testing.T) {
+	if testing.Short() {
+		t.Skip("skipping end-to-end restart test in short mode")
+	}
+
+	sc := &Scenario{
+		Name:  "restart_persist",
+		Cols:  120,
+		Rows:  40,
+		Trust: []string{"persist-target"},
+		Presets: ScenarioPresets{
+			Processes: []ScenarioPreset{{
+				Name: "persist-target",
+				Argv: []string{"persist-target"},
+			}},
+		},
+		Scripts: []ScenarioScript{{
+			Name: "persist-target",
+			Body: "#!/bin/sh\necho RESTORED\nsleep 30\n",
+		}},
+	}
+	env, childEnv, err := prepareEnv(Options{Scenario: sc})
+	if err != nil {
+		t.Fatalf("prepareEnv: %v", err)
+	}
+	t.Cleanup(func() { _ = os.RemoveAll(env.Root) })
+
+	// ── Session 1 — spawn the process via MCP. ──────────────────
+	s1 := openSession(t, env, childEnv)
+	spawnRaw, err := s1.MCPCall("spawn_process", mustJSON(t, map[string]any{
+		"preset": "persist-target",
+	}))
+	if err != nil {
+		_ = s1.Close()
+		t.Fatalf("spawn_process: %v", err)
+	}
+	var spawned map[string]any
+	if err := json.Unmarshal(spawnRaw, &spawned); err != nil {
+		_ = s1.Close()
+		t.Fatalf("decode spawn: %v", err)
+	}
+	if id, _ := spawned["process_id"].(string); id == "" {
+		_ = s1.Close()
+		t.Fatalf("spawn returned no process_id: %s", string(spawnRaw))
+	}
+
+	if err := waitForListEntry(s1, "persist-target", 3*time.Second); err != nil {
+		_ = s1.Close()
+		t.Fatalf("list_processes (session 1): %v", err)
+	}
+
+	// Verify the on-disk record exists before tearing down.
+	stateFile := filepath.Join(env.DataHome, "patterm", "projects")
+	if entries, err := os.ReadDir(stateFile); err != nil || len(entries) == 0 {
+		_ = s1.Close()
+		t.Fatalf("expected per-project state dir under %s before shutdown: err=%v entries=%v", stateFile, err, entries)
+	}
+
+	if err := s1.Close(); err != nil {
+		t.Fatalf("close session 1: %v", err)
+	}
+
+	// ── Session 2 — same env, same project. The persisted entry
+	// must be replayed and show up in list_processes again. ─────
+	s2 := openSession(t, env, childEnv)
+	t.Cleanup(func() { _ = s2.Close() })
+
+	if err := waitForListEntry(s2, "persist-target", 5*time.Second); err != nil {
+		t.Fatalf("list_processes (session 2): %v", err)
+	}
+
+	// Closing the restored process should also drop it from the
+	// persist store, so a third session starts clean.
+	listRaw, err := s2.MCPCall("list_processes", json.RawMessage(`{}`))
+	if err != nil {
+		t.Fatalf("list_processes: %v", err)
+	}
+	var list []map[string]any
+	if err := json.Unmarshal(listRaw, &list); err != nil {
+		t.Fatalf("decode list: %v", err)
+	}
+	var restoredID string
+	for _, p := range list {
+		if name, _ := p["name"].(string); name == "persist-target" {
+			restoredID, _ = p["process_id"].(string)
+			break
+		}
+	}
+	if restoredID == "" {
+		t.Fatalf("restored process missing id in list: %s", string(listRaw))
+	}
+	if _, err := s2.MCPCall("close_process", mustJSON(t, map[string]any{
+		"process_id": restoredID,
+	})); err != nil {
+		t.Fatalf("close_process: %v", err)
+	}
+
+	if err := s2.Close(); err != nil {
+		t.Fatalf("close session 2: %v", err)
+	}
+
+	s3 := openSession(t, env, childEnv)
+	t.Cleanup(func() { _ = s3.Close() })
+	listRaw, err = s3.MCPCall("list_processes", json.RawMessage(`{}`))
+	if err != nil {
+		t.Fatalf("list_processes (session 3): %v", err)
+	}
+	if err := json.Unmarshal(listRaw, &list); err != nil {
+		t.Fatalf("decode list 3: %v", err)
+	}
+	for _, p := range list {
+		if name, _ := p["name"].(string); name == "persist-target" {
+			t.Fatalf("closed process re-appeared in session 3: %s", string(listRaw))
+		}
+	}
+}
+
+// openSession spawns one patterm process against the supplied env and
+// blocks until its MCP socket is ready. Mirrors NewCLI but skips
+// prepareEnv so multiple sessions can share the same XDG dirs.
+func openSession(t *testing.T, env *testEnv, childEnv []string) *Session {
+	t.Helper()
+	em, err := vt.NewGhosttyEmulator(env.Cols, env.Rows)
+	if err != nil {
+		t.Fatalf("vt emulator: %v", err)
+	}
+	p, err := pkgpty.Start([]string{env.PattermBin, "--in-process", "--project", env.ProjectDir}, childEnv, "", env.Cols, env.Rows)
+	if err != nil {
+		_ = em.Close()
+		t.Fatalf("pty start: %v", err)
+	}
+	em.OnWritePTY(func(b []byte) { _, _ = p.Write(b) })
+	s := &Session{pty: p, em: em, env: env, readerDone: make(chan struct{})}
+	go s.readLoop()
+	if err := s.bootstrapMCP(3 * time.Second); err != nil {
+		_ = s.Close()
+		t.Fatalf("mcp bootstrap: %v", err)
+	}
+	return s
+}
+
+func waitForListEntry(s *Session, name string, timeout time.Duration) error {
+	deadline := time.Now().Add(timeout)
+	for time.Now().Before(deadline) {
+		raw, err := s.MCPCall("list_processes", json.RawMessage(`{}`))
+		if err == nil {
+			var list []map[string]any
+			if err := json.Unmarshal(raw, &list); err == nil {
+				for _, p := range list {
+					if n, _ := p["name"].(string); n == name {
+						return nil
+					}
+				}
+			}
+		}
+		time.Sleep(50 * time.Millisecond)
+	}
+	return fmt.Errorf("process %q never appeared in list_processes within %s", name, timeout)
+}
+
+func mustJSON(t *testing.T, v any) json.RawMessage {
+	t.Helper()
+	b, err := json.Marshal(v)
+	if err != nil {
+		t.Fatalf("marshal: %v", err)
+	}
+	return b
+}

internal/harness/runner.go

@@ -5,6 +5,7 @@ import (
 	"fmt"
 	"regexp"
 	"strings"
+	"time"
 )
 
 type Event struct {
@@ -175,6 +176,41 @@ func runStep(s *Session, step Step, results map[string]json.RawMessage) error {
 			return fmt.Errorf("no saved result %q", step.From)
 		}
 		return assertJSONValue(raw, step.Path, step.Equals, step.Contains, step.AllowSubstring)
+	case "wait_until_mcp":
+		// Poll an MCP method until the assertion at Path holds (or
+		// Contains substring matches), or TimeoutMS elapses. Used by the
+		// idle-detection scenarios to wait for a child's idle_state to
+		// reach a target value without sprinkling sleeps.
+		params, perr := resolveParams(step.Params, results)
+		if perr != nil {
+			return perr
+		}
+		deadline := time.Now().Add(timeoutMS(step.TimeoutMS))
+		var lastRaw json.RawMessage
+		var lastErr error
+		for {
+			raw, err := s.MCPCall(step.Method, params)
+			if err == nil {
+				if aerr := assertJSONValue(raw, step.Path, step.Equals, step.Contains, step.AllowSubstring); aerr == nil {
+					if step.SaveAs != "" {
+						results[step.SaveAs] = raw
+					}
+					return nil
+				} else {
+					lastErr = aerr
+					lastRaw = raw
+				}
+			} else {
+				lastErr = err
+			}
+			if time.Now().After(deadline) {
+				if lastErr != nil {
+					return fmt.Errorf("wait_until_mcp timeout: %w (last response: %s)", lastErr, string(lastRaw))
+				}
+				return fmt.Errorf("wait_until_mcp timeout (no successful call)")
+			}
+			time.Sleep(100 * time.Millisecond)
+		}
 	}
 	return fmt.Errorf("unknown step type %q", step.Type)
 }

internal/harness/scenario.go

@@ -25,11 +25,23 @@ type ScenarioPresets struct {
 }
 
 type ScenarioPreset struct {
-	Name       string            `json:"name"`
-	Argv       []string          `json:"argv"`
-	Env        map[string]string `json:"env,omitempty"`
-	WorkingDir string            `json:"working_dir,omitempty"`
-	Shell      bool              `json:"shell,omitempty"`
+	Name           string                  `json:"name"`
+	Argv           []string                `json:"argv"`
+	Env            map[string]string       `json:"env,omitempty"`
+	WorkingDir     string                  `json:"working_dir,omitempty"`
+	Shell          bool                    `json:"shell,omitempty"`
+	IdleDetection  *ScenarioIdleDetection  `json:"idle_detection,omitempty"`
+}
+
+// ScenarioIdleDetection mirrors preset.IdleDetection so scenarios can
+// configure per-strategy idle detection for fake agent presets.
+type ScenarioIdleDetection struct {
+	Strategy           string            `json:"strategy,omitempty"`
+	IdleThresholdMS    int               `json:"idle_threshold_ms,omitempty"`
+	TitleStatusMap     map[string]string `json:"title_status_map,omitempty"`
+	PermissionPatterns []string          `json:"permission_patterns,omitempty"`
+	ThinkingPatterns   []string          `json:"thinking_patterns,omitempty"`
+	ErrorPatterns      []string          `json:"error_patterns,omitempty"`
 }
 
 type ScenarioScript struct {

internal/harness/scenarios/chrome_survives_origin_mode.json

@@ -0,0 +1,32 @@
+{
+  "name": "chrome_survives_origin_mode",
+  "cols": 120,
+  "rows": 40,
+  "scripts": [
+    {
+      "name": "origin-mode",
+      "body": "#!/bin/sh\n# Child TUIs are allowed to use DEC origin mode internally, but the\n# host chrome must never inherit it. If CSI ? 6 h reaches the real\n# terminal, patterm's absolute CUPs for the tab bar/status/sidebar are\n# interpreted relative to the child scroll region and chrome appears\n# inside the viewport.\nprintf 'ORIGIN READY\\n'\nsleep 0.1\nprintf '\\033[5;20r'\nprintf '\\033[?6h'\nprintf '\\033[1;1HORIGIN MODE ACTIVE\\n'\nsleep 0.2\nprintf 'ORIGIN DONE\\n'\nsleep 5\n"
+    }
+  ],
+  "steps": [
+    {
+      "type": "mcp_call",
+      "method": "spawn_process",
+      "params": { "kind": "command", "argv": ["origin-mode"], "name": "origin-mode" }
+    },
+    { "type": "wait_text", "contains": "ORIGIN DONE", "timeout_ms": 5000 },
+    { "type": "wait_stable", "timeout_ms": 2000 },
+    { "type": "assert_contains", "contains": "+ new" },
+    { "type": "assert_contains", "contains": "Processes" },
+    { "type": "assert_contains", "contains": "Agent Tree" },
+    { "type": "assert_contains", "contains": "Scratchpads" },
+    {
+      "type": "assert_regex",
+      "regex": "(?m)^[^\\n]*\\+ new[^\\n]*Processes[^\\n]*$"
+    },
+    {
+      "type": "assert_regex",
+      "regex": "(?m)^origin-mode · you have control[^\\n]*Ctrl-K · palette[^\\n]*$"
+    }
+  ]
+}

internal/harness/scenarios/error_flash_preserves_focused_pane.json

@@ -0,0 +1,37 @@
+{
+  "name": "error_flash_preserves_focused_pane",
+  "presets": {
+    "processes": [
+      {
+        "name": "steady",
+        "argv": ["sh", "-lc", "printf 'STEADY READY\\n'; sleep 5"]
+      }
+    ]
+  },
+  "trust": ["steady"],
+  "steps": [
+    {
+      "type": "mcp_call",
+      "method": "spawn_process",
+      "params": {"kind": "command", "preset": "steady", "name": "steady"},
+      "save_as": "proc"
+    },
+    { "type": "wait_text", "contains": "STEADY READY", "timeout_ms": 5000 },
+    { "type": "send_chord", "chord": "ctrl-k" },
+    { "type": "send_text", "text": "Open Settings" },
+    { "type": "send_chord", "chord": "enter" },
+    { "type": "send_chord", "chord": "enter" },
+    { "type": "send_chord", "chord": "ctrl-n" },
+    { "type": "send_chord", "chord": "ctrl-n" },
+    { "type": "send_chord", "chord": "ctrl-n" },
+    { "type": "send_chord", "chord": "ctrl-n" },
+    { "type": "send_chord", "chord": "ctrl-n" },
+    { "type": "send_chord", "chord": "ctrl-n" },
+    { "type": "send_chord", "chord": "ctrl-n" },
+    { "type": "send_chord", "chord": "enter" },
+    { "type": "wait_text", "contains": "no active top-level agent to summarize", "timeout_ms": 5000 },
+    { "type": "wait_text", "contains": "STEADY READY", "timeout_ms": 5000 },
+    { "type": "assert_contains", "contains": "STEADY READY" },
+    { "type": "assert_not_contains", "contains": "Press Ctrl-K to spawn an agent or process" }
+  ]
+}

internal/harness/scenarios/idle_osc_title_stability.json

@@ -0,0 +1,44 @@
+{
+  "name": "idle_osc_title_stability",
+  "presets": {
+    "processes": [
+      {
+        "name": "titler",
+        "argv": [
+          "sh",
+          "-lc",
+          "i=0; while [ $i -lt 6 ]; do printf '\\033]2;step %d\\007' $i; i=$((i+1)); sleep 0.2; done; sleep 60"
+        ],
+        "idle_detection": {
+          "strategy": "osc_title_stability",
+          "idle_threshold_ms": 1000
+        }
+      }
+    ]
+  },
+  "trust": ["titler"],
+  "steps": [
+    {
+      "type": "mcp_call",
+      "method": "spawn_process",
+      "params": {"kind": "command", "preset": "titler", "name": "titler"},
+      "save_as": "proc"
+    },
+    {
+      "type": "wait_until_mcp",
+      "method": "get_process_status",
+      "params": {"process_id": "{{proc.process_id}}"},
+      "path": "idle_state",
+      "equals": "working",
+      "timeout_ms": 3000
+    },
+    {
+      "type": "wait_until_mcp",
+      "method": "get_process_status",
+      "params": {"process_id": "{{proc.process_id}}"},
+      "path": "idle_state",
+      "equals": "idle",
+      "timeout_ms": 4000
+    }
+  ]
+}

internal/harness/scenarios/idle_osc_title_status.json

@@ -0,0 +1,48 @@
+{
+  "name": "idle_osc_title_status",
+  "presets": {
+    "processes": [
+      {
+        "name": "geminilike",
+        "argv": [
+          "sh",
+          "-lc",
+          "printf '\\033]2;Thinking\\007'; sleep 1; printf '\\033]2;Permission required\\007'; sleep 60"
+        ],
+        "idle_detection": {
+          "strategy": "osc_title_status",
+          "idle_threshold_ms": 1000,
+          "title_status_map": {
+            "thinking": "thinking",
+            "permission": "permission"
+          }
+        }
+      }
+    ]
+  },
+  "trust": ["geminilike"],
+  "steps": [
+    {
+      "type": "mcp_call",
+      "method": "spawn_process",
+      "params": {"kind": "command", "preset": "geminilike", "name": "geminilike"},
+      "save_as": "proc"
+    },
+    {
+      "type": "wait_until_mcp",
+      "method": "get_process_status",
+      "params": {"process_id": "{{proc.process_id}}"},
+      "path": "idle_state",
+      "equals": "thinking",
+      "timeout_ms": 3000
+    },
+    {
+      "type": "wait_until_mcp",
+      "method": "get_process_status",
+      "params": {"process_id": "{{proc.process_id}}"},
+      "path": "idle_state",
+      "equals": "permission",
+      "timeout_ms": 4000
+    }
+  ]
+}

internal/harness/scenarios/idle_output_activity.json

@@ -0,0 +1,44 @@
+{
+  "name": "idle_output_activity",
+  "presets": {
+    "processes": [
+      {
+        "name": "blinker",
+        "argv": ["sh", "-lc", "echo step1; sleep 3; echo step2; sleep 60"],
+        "idle_detection": {
+          "strategy": "output_activity",
+          "idle_threshold_ms": 1000
+        }
+      }
+    ]
+  },
+  "trust": ["blinker"],
+  "steps": [
+    {
+      "type": "mcp_call",
+      "method": "spawn_process",
+      "params": {
+        "kind": "command",
+        "preset": "blinker",
+        "name": "blinker"
+      },
+      "save_as": "proc"
+    },
+    {
+      "type": "wait_until_mcp",
+      "method": "get_process_status",
+      "params": {"process_id": "{{proc.process_id}}"},
+      "path": "idle_state",
+      "equals": "working",
+      "timeout_ms": 4000
+    },
+    {
+      "type": "wait_until_mcp",
+      "method": "get_process_status",
+      "params": {"process_id": "{{proc.process_id}}"},
+      "path": "idle_state",
+      "equals": "idle",
+      "timeout_ms": 4000
+    }
+  ]
+}

internal/harness/scenarios/idle_regex_promote.json

@@ -0,0 +1,33 @@
+{
+  "name": "idle_regex_promote",
+  "presets": {
+    "processes": [
+      {
+        "name": "approver",
+        "argv": ["sh", "-lc", "echo 'Do you want to proceed?'; sleep 60"],
+        "idle_detection": {
+          "strategy": "output_activity",
+          "idle_threshold_ms": 500,
+          "permission_patterns": ["Do you want to proceed\\?"]
+        }
+      }
+    ]
+  },
+  "trust": ["approver"],
+  "steps": [
+    {
+      "type": "mcp_call",
+      "method": "spawn_process",
+      "params": {"kind": "command", "preset": "approver", "name": "approver"},
+      "save_as": "proc"
+    },
+    {
+      "type": "wait_until_mcp",
+      "method": "get_process_status",
+      "params": {"process_id": "{{proc.process_id}}"},
+      "path": "idle_state",
+      "equals": "permission",
+      "timeout_ms": 4000
+    }
+  ]
+}

internal/harness/scenarios/idle_screen_permission_prompt.json

@@ -0,0 +1,37 @@
+{
+  "name": "idle_screen_permission_prompt",
+  "presets": {
+    "processes": [
+      {
+        "name": "screen-permission",
+        "argv": [
+          "sh",
+          "-lc",
+          "printf '\\033[2J\\033[HCalling patterm...\\n\\nTool use\\n\\nDo you want to proceed?\\n  1. Yes\\n'; i=0; while [ $i -lt 300 ]; do printf '\\033[HCalling patterm... %03d' $i; i=$((i+1)); done; sleep 60"
+        ],
+        "idle_detection": {
+          "strategy": "output_activity",
+          "idle_threshold_ms": 500,
+          "permission_patterns": ["Do you want to proceed\\?"]
+        }
+      }
+    ]
+  },
+  "trust": ["screen-permission"],
+  "steps": [
+    {
+      "type": "mcp_call",
+      "method": "spawn_process",
+      "params": {"kind": "command", "preset": "screen-permission", "name": "screen-permission"},
+      "save_as": "proc"
+    },
+    {
+      "type": "wait_until_mcp",
+      "method": "get_process_status",
+      "params": {"process_id": "{{proc.process_id}}"},
+      "path": "idle_state",
+      "equals": "permission",
+      "timeout_ms": 4000
+    }
+  ]
+}

internal/harness/scenarios/palette_over_scratchpad.json

@@ -0,0 +1,28 @@
+{
+  "name": "palette_over_scratchpad",
+  "cols": 120,
+  "rows": 30,
+  "steps": [
+    {
+      "type": "mcp_call",
+      "method": "scratchpad_write",
+      "params": { "name": "pad-marker.md", "content": "# Pad Heading\n\nzealot-marker body line" }
+    },
+    { "type": "wait_stable", "timeout_ms": 2000 },
+    { "type": "send_chord", "chord": "ctrl-s" },
+    { "type": "wait_text", "contains": "zealot-marker", "timeout_ms": 5000 },
+    { "type": "assert_contains", "contains": "Pad Heading" },
+
+    { "type": "send_chord", "chord": "ctrl-k" },
+    { "type": "wait_stable", "timeout_ms": 2000 },
+    { "type": "send_text", "text": "quit" },
+    { "type": "wait_text", "contains": "quit", "timeout_ms": 5000 },
+    { "type": "assert_contains", "contains": "quit" },
+
+    { "type": "send_chord", "chord": "escape" },
+    { "type": "wait_text", "contains": "zealot-marker", "timeout_ms": 5000 },
+    { "type": "assert_contains", "contains": "Pad Heading" },
+    { "type": "assert_contains", "contains": "zealot-marker" },
+    { "type": "assert_not_contains", "contains": "quit" }
+  ]
+}

internal/harness/scenarios/rename_process_via_palette.json

@@ -0,0 +1,31 @@
+{
+  "name": "rename_process_via_palette",
+  "scripts": [
+    {
+      "name": "renamed-loop",
+      "body": "#!/bin/sh\necho RENAMED READY\nsleep 5\n"
+    }
+  ],
+  "steps": [
+    {
+      "type": "mcp_call",
+      "method": "spawn_process",
+      "params": { "kind": "command", "argv": ["renamed-loop"], "name": "original" }
+    },
+    { "type": "wait_text", "contains": "RENAMED READY", "timeout_ms": 5000 },
+    { "type": "send_chord", "chord": "ctrl-k" },
+    { "type": "send_text", "text": "Rename process" },
+    { "type": "send_chord", "chord": "enter" },
+    { "type": "wait_text", "contains": "process: original", "timeout_ms": 3000 },
+    { "type": "send_chord", "chord": "ctrl-u" },
+    { "type": "send_text", "text": "renamed-pane" },
+    { "type": "send_chord", "chord": "enter" },
+    { "type": "wait_stable", "timeout_ms": 2000 },
+    {
+      "type": "assert_mcp",
+      "method": "get_project_status",
+      "path": "processes.0.name",
+      "equals": "renamed-pane"
+    }
+  ]
+}

internal/harness/scenarios/restart_exited_process_from_sidebar.json

@@ -0,0 +1,33 @@
+{
+  "name": "restart_exited_process_from_sidebar",
+  "cols": 120,
+  "rows": 40,
+  "scripts": [
+    {
+      "name": "quick-shell",
+      "body": "#!/bin/sh\ncount_file=\"$XDG_RUNTIME_DIR/quick-shell-count\"\nif [ -f \"$count_file\" ]; then\n  n=$(cat \"$count_file\")\nelse\n  n=0\nfi\nn=$((n + 1))\nprintf '%s\\n' \"$n\" > \"$count_file\"\nprintf 'QUICK RUN %s\\n' \"$n\"\n"
+    }
+  ],
+  "steps": [
+    {
+      "type": "mcp_call",
+      "method": "spawn_process",
+      "params": { "kind": "command", "argv": ["quick-shell"], "name": "quick-shell" }
+    },
+    { "type": "wait_text", "contains": "QUICK RUN 1", "timeout_ms": 5000 },
+    { "type": "wait_stable", "timeout_ms": 2000 },
+    { "type": "assert_contains", "contains": "○ quick-shell" },
+    { "type": "send_text", "text": "\u0017" },
+    { "type": "wait_stable", "timeout_ms": 2000 },
+    { "type": "assert_contains", "contains": "quick-shell · you have control" },
+    { "type": "mark_raw", "save_as": "before_restart" },
+    { "type": "send_text", "text": "\u0012" },
+    { "type": "wait_text", "contains": "QUICK RUN 2", "timeout_ms": 5000 },
+    {
+      "type": "assert_raw_since_regex",
+      "from": "before_restart",
+      "regex": "QUICK RUN 2",
+      "timeout_ms": 2000
+    }
+  ]
+}

internal/harness/scenarios/restart_process_keeps_chrome.json

@@ -0,0 +1,32 @@
+{
+  "name": "restart_process_keeps_chrome",
+  "cols": 120,
+  "rows": 40,
+  "scripts": [
+    {
+      "name": "slow-restart",
+      "body": "#!/bin/sh\ncount_file=\"$XDG_RUNTIME_DIR/slow-restart-count\"\nif [ -f \"$count_file\" ]; then\n  n=$(cat \"$count_file\")\nelse\n  n=0\nfi\nn=$((n + 1))\nprintf '%s\\n' \"$n\" > \"$count_file\"\nprintf 'SLOW READY %s\\n' \"$n\"\ntrap 'sleep 3; exit 0' TERM\nwhile true; do sleep 1; done\n"
+    }
+  ],
+  "steps": [
+    {
+      "type": "mcp_call",
+      "method": "spawn_process",
+      "params": { "kind": "command", "argv": ["slow-restart"], "name": "slow-restart" },
+      "save_as": "spawned"
+    },
+    {
+      "type": "mcp_call",
+      "method": "select_process",
+      "params": { "process_id": "{{spawned.process_id}}" }
+    },
+    { "type": "wait_text", "contains": "SLOW READY 1", "timeout_ms": 5000 },
+    { "type": "wait_stable", "timeout_ms": 2000 },
+    { "type": "assert_contains", "contains": "Processes" },
+    { "type": "send_text", "text": "\u000brestart\r" },
+    { "type": "wait_stable", "timeout_ms": 2000 },
+    { "type": "assert_contains", "contains": "Processes" },
+    { "type": "assert_contains", "contains": "slow-restart" },
+    { "type": "wait_text", "contains": "SLOW READY 2", "timeout_ms": 7000 }
+  ]
+}

internal/harness/scenarios/scratchpad_focus.json

@@ -0,0 +1,18 @@
+{
+  "name": "scratchpad_focus",
+  "cols": 120,
+  "rows": 40,
+  "steps": [
+    {
+      "type": "mcp_call",
+      "method": "scratchpad_write",
+      "params": { "name": "notes.md", "content": "# Heading One\n\n- item alpha\n- item beta\n\nhello scratchpad" }
+    },
+    { "type": "wait_stable", "timeout_ms": 2000 },
+    { "type": "assert_contains", "contains": "notes.md" },
+    { "type": "send_chord", "chord": "ctrl-s" },
+    { "type": "wait_text", "contains": "hello scratchpad", "timeout_ms": 5000 },
+    { "type": "assert_contains", "contains": "Heading One" },
+    { "type": "assert_contains", "contains": "item alpha" }
+  ]
+}

internal/harness/scenarios/scratchpad_scroll.json

@@ -0,0 +1,40 @@
+{
+  "name": "scratchpad_scroll",
+  "cols": 120,
+  "rows": 20,
+  "steps": [
+    {
+      "type": "mcp_call",
+      "method": "scratchpad_write",
+      "params": {
+        "name": "long.md",
+        "content": "# Long pad\n\nline-01\nline-02\nline-03\nline-04\nline-05\nline-06\nline-07\nline-08\nline-09\nline-10\nline-11\nline-12\nline-13\nline-14\nline-15\nline-16\nline-17\nline-18\nline-19\nline-20\nline-21\nline-22\nline-23\nline-24\nline-25\nline-26\nline-27\nline-28\nline-29\nline-30\nfinal-marker"
+      }
+    },
+    { "type": "wait_stable", "timeout_ms": 2000 },
+    { "type": "send_chord", "chord": "ctrl-s" },
+    { "type": "wait_text", "contains": "line-01", "timeout_ms": 5000 },
+    { "type": "assert_not_contains", "contains": "final-marker" },
+    { "type": "send_chord", "chord": "wheel-down" },
+    { "type": "send_chord", "chord": "wheel-down" },
+    { "type": "send_chord", "chord": "wheel-down" },
+    { "type": "send_chord", "chord": "wheel-down" },
+    { "type": "send_chord", "chord": "wheel-down" },
+    { "type": "send_chord", "chord": "wheel-down" },
+    { "type": "send_chord", "chord": "wheel-down" },
+    { "type": "wait_text", "contains": "final-marker", "timeout_ms": 5000 },
+    { "type": "assert_contains", "contains": "final-marker" },
+    { "type": "send_chord", "chord": "wheel-up" },
+    { "type": "send_chord", "chord": "wheel-up" },
+    { "type": "send_chord", "chord": "wheel-up" },
+    { "type": "send_chord", "chord": "wheel-up" },
+    { "type": "send_chord", "chord": "wheel-up" },
+    { "type": "send_chord", "chord": "wheel-up" },
+    { "type": "send_chord", "chord": "wheel-up" },
+    { "type": "send_chord", "chord": "wheel-up" },
+    { "type": "send_chord", "chord": "wheel-up" },
+    { "type": "send_chord", "chord": "wheel-up" },
+    { "type": "wait_text", "contains": "line-01", "timeout_ms": 5000 },
+    { "type": "assert_contains", "contains": "line-01" }
+  ]
+}

internal/harness/scenarios/sidebar_survives_linefeed_scroll.json

@@ -0,0 +1,34 @@
+{
+  "name": "sidebar_survives_linefeed_scroll",
+  "cols": 120,
+  "rows": 40,
+  "scripts": [
+    {
+      "name": "linefeed-scroll",
+      "body": "#!/bin/sh\n# Plain LF at the bottom of the child viewport scrolls the host's\n# DECSTBM region. Because that region spans every column, enough LFs\n# drag the sidebar border and section labels out of the visible region\n# unless patterm invalidates and repaints the sidebar cache.\ni=0\nwhile [ $i -lt 12 ]; do\n  printf 'warmup %02d\\n' \"$i\"\n  i=$((i + 1))\n  sleep 0.05\ndone\nprintf 'LINEFEED READY\\n'\nIFS= read -r _\nprintf '\\033[1;36r'\nprintf '\\033[36;1H'\ni=0\nwhile [ $i -lt 45 ]; do\n  printf 'scroll line %02d\\n' \"$i\"\n  i=$((i + 1))\ndone\nprintf 'LINEFEED DONE\\n'\nsleep 5\n"
+    }
+  ],
+  "steps": [
+    {
+      "type": "mcp_call",
+      "method": "spawn_process",
+      "params": { "kind": "command", "argv": ["linefeed-scroll"], "name": "linefeed-scroll" }
+    },
+    { "type": "wait_text", "contains": "LINEFEED READY", "timeout_ms": 5000 },
+    { "type": "wait_stable", "timeout_ms": 2000 },
+    { "type": "mark_raw", "save_as": "before_scroll" },
+    { "type": "send_chord", "chord": "enter" },
+    { "type": "wait_text", "contains": "LINEFEED DONE", "timeout_ms": 5000 },
+    { "type": "wait_stable", "timeout_ms": 2000 },
+    {
+      "type": "assert_raw_since_regex",
+      "from": "before_scroll",
+      "regex": "LINEFEED DONE",
+      "timeout_ms": 2000
+    },
+    { "type": "assert_contains", "contains": "Processes" },
+    { "type": "assert_contains", "contains": "Agent Tree" },
+    { "type": "assert_contains", "contains": "Scratchpads" },
+    { "type": "assert_contains", "contains": "● linefeed-scroll" }
+  ]
+}

internal/harness/scenarios/timer_cancel.json

@@ -0,0 +1,44 @@
+{
+  "name": "timer_cancel",
+  "presets": {
+    "processes": [
+      {
+        "name": "echoer",
+        "argv": ["sh", "-lc", "while read line; do echo \"saw:$line\"; done"]
+      }
+    ]
+  },
+  "trust": ["echoer"],
+  "steps": [
+    {
+      "type": "mcp_call",
+      "method": "spawn_process",
+      "params": {"kind": "command", "preset": "echoer", "name": "echoer"},
+      "save_as": "proc"
+    },
+    { "type": "wait_stable", "timeout_ms": 1500 },
+    {
+      "type": "mcp_call",
+      "method": "timer_set",
+      "params": {"seconds": 1, "body": "should-not-arrive", "owner_process_id": "{{proc.process_id}}"},
+      "save_as": "tmr"
+    },
+    {
+      "type": "mcp_call",
+      "method": "timer_cancel",
+      "params": {"timer_id": "{{tmr.timer_id}}"}
+    },
+    {
+      "type": "mcp_call",
+      "method": "timer_list",
+      "params": {"owner_process_id": "{{proc.process_id}}"},
+      "save_as": "listed"
+    },
+    {
+      "type": "assert_saved",
+      "from": "listed",
+      "path": "",
+      "equals": []
+    }
+  ]
+}

internal/harness/scenarios/timer_idle_all_already_satisfied.json

@@ -0,0 +1,48 @@
+{
+  "name": "timer_idle_all_already_satisfied",
+  "presets": {
+    "processes": [
+      {
+        "name": "quiet",
+        "argv": ["sh", "-lc", "echo ready; sleep 60"],
+        "idle_detection": {
+          "strategy": "output_activity",
+          "idle_threshold_ms": 500
+        }
+      }
+    ]
+  },
+  "trust": ["quiet"],
+  "steps": [
+    {
+      "type": "mcp_call",
+      "method": "spawn_process",
+      "params": {"kind": "command", "preset": "quiet", "name": "quiet"},
+      "save_as": "proc"
+    },
+    {
+      "type": "wait_until_mcp",
+      "method": "get_process_status",
+      "params": {"process_id": "{{proc.process_id}}"},
+      "path": "idle_state",
+      "equals": "idle",
+      "timeout_ms": 4000
+    },
+    {
+      "type": "mcp_call",
+      "method": "timer_fire_when_idle_all",
+      "params": {
+        "watched": ["{{proc.process_id}}"],
+        "body": "all-idle",
+        "owner_process_id": "{{proc.process_id}}"
+      },
+      "save_as": "resp"
+    },
+    {
+      "type": "assert_saved",
+      "from": "resp",
+      "path": "status",
+      "equals": "already_satisfied"
+    }
+  ]
+}

internal/harness/scenarios/timer_idle_all_pending.json

@@ -0,0 +1,89 @@
+{
+  "name": "timer_idle_all_pending",
+  "presets": {
+    "processes": [
+      {
+        "name": "echoer",
+        "argv": ["sh", "-lc", "while read line; do echo \"saw:$line\"; done"]
+      },
+      {
+        "name": "quiet",
+        "argv": ["sh", "-lc", "echo ready; sleep 60"],
+        "idle_detection": {
+          "strategy": "output_activity",
+          "idle_threshold_ms": 500
+        }
+      },
+      {
+        "name": "busy",
+        "argv": ["sh", "-lc", "for i in 1 2 3 4 5; do echo tick $i; sleep 0.2; done; sleep 60"],
+        "idle_detection": {
+          "strategy": "output_activity",
+          "idle_threshold_ms": 500
+        }
+      }
+    ]
+  },
+  "trust": ["echoer", "quiet", "busy"],
+  "steps": [
+    {
+      "type": "mcp_call",
+      "method": "spawn_process",
+      "params": {"kind": "command", "preset": "echoer", "name": "echoer"},
+      "save_as": "owner"
+    },
+    {
+      "type": "mcp_call",
+      "method": "spawn_process",
+      "params": {"kind": "command", "preset": "quiet", "name": "quiet"},
+      "save_as": "q"
+    },
+    {
+      "type": "mcp_call",
+      "method": "spawn_process",
+      "params": {"kind": "command", "preset": "busy", "name": "busy"},
+      "save_as": "b"
+    },
+    {
+      "type": "wait_until_mcp",
+      "method": "get_process_status",
+      "params": {"process_id": "{{q.process_id}}"},
+      "path": "idle_state",
+      "equals": "idle",
+      "timeout_ms": 3000
+    },
+    {
+      "type": "wait_until_mcp",
+      "method": "get_process_status",
+      "params": {"process_id": "{{b.process_id}}"},
+      "path": "idle_state",
+      "equals": "working",
+      "timeout_ms": 3000
+    },
+    {
+      "type": "mcp_call",
+      "method": "timer_fire_when_idle_all",
+      "params": {
+        "watched": ["{{q.process_id}}", "{{b.process_id}}"],
+        "body": "all-idle",
+        "owner_process_id": "{{owner.process_id}}"
+      },
+      "save_as": "resp"
+    },
+    {
+      "type": "assert_saved",
+      "from": "resp",
+      "path": "status",
+      "equals": "pending"
+    },
+    {
+      "type": "wait_until_mcp",
+      "method": "get_process_output",
+      "params": {"process_id": "{{owner.process_id}}", "mode": "grid"},
+      "path": "content",
+      "contains": "saw:all-idle",
+      "allow_substring": true,
+      "timeout_ms": 6000
+    }
+  ]
+}

internal/harness/scenarios/timer_idle_any_fires_on_transition.json

@@ -0,0 +1,67 @@
+{
+  "name": "timer_idle_any_fires_on_transition",
+  "presets": {
+    "processes": [
+      {
+        "name": "echoer",
+        "argv": ["sh", "-lc", "while read line; do echo \"saw:$line\"; done"]
+      },
+      {
+        "name": "busy",
+        "argv": ["sh", "-lc", "for i in 1 2 3 4 5; do echo tick $i; sleep 0.2; done; sleep 60"],
+        "idle_detection": {
+          "strategy": "output_activity",
+          "idle_threshold_ms": 500
+        }
+      }
+    ]
+  },
+  "trust": ["echoer", "busy"],
+  "steps": [
+    {
+      "type": "mcp_call",
+      "method": "spawn_process",
+      "params": {"kind": "command", "preset": "echoer", "name": "echoer"},
+      "save_as": "owner"
+    },
+    {
+      "type": "mcp_call",
+      "method": "spawn_process",
+      "params": {"kind": "command", "preset": "busy", "name": "busy"},
+      "save_as": "watch"
+    },
+    {
+      "type": "wait_until_mcp",
+      "method": "get_process_status",
+      "params": {"process_id": "{{watch.process_id}}"},
+      "path": "idle_state",
+      "equals": "working",
+      "timeout_ms": 3000
+    },
+    {
+      "type": "mcp_call",
+      "method": "timer_fire_when_idle_any",
+      "params": {
+        "watched": ["{{watch.process_id}}"],
+        "body": "any-idle",
+        "owner_process_id": "{{owner.process_id}}"
+      },
+      "save_as": "resp"
+    },
+    {
+      "type": "assert_saved",
+      "from": "resp",
+      "path": "status",
+      "equals": "pending"
+    },
+    {
+      "type": "wait_until_mcp",
+      "method": "get_process_output",
+      "params": {"process_id": "{{owner.process_id}}", "mode": "grid"},
+      "path": "content",
+      "contains": "saw:any-idle",
+      "allow_substring": true,
+      "timeout_ms": 6000
+    }
+  ]
+}

internal/harness/scenarios/timer_pause_resume.json

@@ -0,0 +1,62 @@
+{
+  "name": "timer_pause_resume",
+  "presets": {
+    "processes": [
+      {
+        "name": "echoer",
+        "argv": ["sh", "-lc", "while read line; do echo \"saw:$line\"; done"]
+      }
+    ]
+  },
+  "trust": ["echoer"],
+  "steps": [
+    {
+      "type": "mcp_call",
+      "method": "spawn_process",
+      "params": {"kind": "command", "preset": "echoer", "name": "echoer"},
+      "save_as": "proc"
+    },
+    { "type": "wait_stable", "timeout_ms": 1500 },
+    {
+      "type": "mcp_call",
+      "method": "timer_set",
+      "params": {
+        "seconds": 1,
+        "body": "after-resume",
+        "owner_process_id": "{{proc.process_id}}"
+      },
+      "save_as": "tmr"
+    },
+    {
+      "type": "mcp_call",
+      "method": "timer_pause",
+      "params": {"timer_id": "{{tmr.timer_id}}"}
+    },
+    {
+      "type": "mcp_call",
+      "method": "timer_list",
+      "params": {"owner_process_id": "{{proc.process_id}}"},
+      "save_as": "listed"
+    },
+    {
+      "type": "assert_saved",
+      "from": "listed",
+      "path": "0.status",
+      "equals": "paused"
+    },
+    {
+      "type": "mcp_call",
+      "method": "timer_resume",
+      "params": {"timer_id": "{{tmr.timer_id}}"}
+    },
+    {
+      "type": "wait_until_mcp",
+      "method": "get_process_output",
+      "params": {"process_id": "{{proc.process_id}}", "mode": "grid"},
+      "path": "content",
+      "contains": "saw:after-resume",
+      "allow_substring": true,
+      "timeout_ms": 5000
+    }
+  ]
+}

internal/harness/scenarios/timer_set_delivers.json

@@ -0,0 +1,40 @@
+{
+  "name": "timer_set_delivers",
+  "presets": {
+    "processes": [
+      {
+        "name": "echoer",
+        "argv": ["sh", "-lc", "while read line; do echo \"saw:$line\"; done"]
+      }
+    ]
+  },
+  "trust": ["echoer"],
+  "steps": [
+    {
+      "type": "mcp_call",
+      "method": "spawn_process",
+      "params": {"kind": "command", "preset": "echoer", "name": "echoer"},
+      "save_as": "proc"
+    },
+    { "type": "wait_stable", "timeout_ms": 1500 },
+    {
+      "type": "mcp_call",
+      "method": "timer_set",
+      "params": {
+        "seconds": 0.5,
+        "body": "hello-from-timer",
+        "owner_process_id": "{{proc.process_id}}"
+      },
+      "save_as": "tmr"
+    },
+    {
+      "type": "wait_until_mcp",
+      "method": "get_process_output",
+      "params": {"process_id": "{{proc.process_id}}", "mode": "grid"},
+      "path": "content",
+      "contains": "saw:hello-from-timer",
+      "allow_substring": true,
+      "timeout_ms": 5000
+    }
+  ]
+}

internal/harness/scenarios/toast_dismiss.json

@@ -0,0 +1,32 @@
+{
+  "name": "toast_dismiss",
+  "presets": {
+    "processes": [
+      {
+        "name": "steady",
+        "argv": ["sh", "-lc", "printf 'STEADY READY\\n'; sleep 30"]
+      }
+    ]
+  },
+  "trust": ["steady"],
+  "steps": [
+    {
+      "type": "mcp_call",
+      "method": "spawn_process",
+      "params": {"kind": "command", "preset": "steady", "name": "steady"},
+      "save_as": "proc"
+    },
+    { "type": "wait_text", "contains": "STEADY READY", "timeout_ms": 5000 },
+    {
+      "type": "mcp_call",
+      "method": "request_human_attention",
+      "params": {"process_id": "{{proc.process_id}}", "reason": "needs eyes on the deploy"}
+    },
+    { "type": "wait_text", "contains": "needs eyes on the deploy", "timeout_ms": 5000 },
+    { "type": "assert_contains", "contains": "STEADY READY" },
+    { "type": "send_chord", "chord": "ctrl-n" },
+    { "type": "wait_stable", "timeout_ms": 2000 },
+    { "type": "assert_contains", "contains": "STEADY READY" },
+    { "type": "assert_not_contains", "contains": "needs eyes on the deploy" }
+  ]
+}

internal/harness/session.go

@@ -55,7 +55,7 @@ func NewCLI(opts Options) (*Session, error) {
 	if err != nil {
 		return nil, err
 	}
-	p, err := pkgpty.Start([]string{env.PattermBin, "--project", env.ProjectDir}, childEnv, env.Cols, env.Rows)
+	p, err := pkgpty.Start([]string{env.PattermBin, "--in-process", "--project", env.ProjectDir}, childEnv, "", env.Cols, env.Rows)
 	if err != nil {
 		_ = em.Close()
 		return nil, err