Release v0.0.2

Bundles the in-flight work into the second tagged release. See
CHANGELOG.md `[0.0.2] - 2026-05-15` for the full per-change list.
Highlights:

- libghostty-vt was building in zig's silent Debug default, capping
  the full pipeline at 34-63 fps. Makefile now defaults to
  ReleaseFast (.mise.toml pins zig 0.15.2 so the build is
  reproducible). End-to-end pipeline now runs at 930-2030 fps —
  27-32× faster, with 7-16× headroom over a 120 fps target.
- --debug[=DIR] and --profile[=DIR] flags capture full PTY logs,
  pprof data, and per-hot-path metrics (chunks/sec, mean/max
  latencies, cache hit rates) for offline analysis. Nothing
  pollutes stdout/stderr.
- ASCII-video benchmark suite (8-colour / truecolor / Bad-Apple
  patterns at 30/60/120 fps) plus a renderer microbenchmark set
  for stable A/B comparisons across changes.
- Click-and-drag text selection from alt-screen TUIs (codex) now
  works — host mouse mode follows the focused child's screen side
  instead of being permanently armed.
- Long claude session resume + codex steady-state rendering pay
  less per chunk: drawSidebar deferred to the chrome ticker,
  emulator.Title CGO poll gated on a containsOSC scan.
- Vendor-TUI orientation: MCP initialize.instructions, the
  spawn_agent tool description, and help('spawning') all spell
  out the anti-patterns (shell-out, perl-into-socket) that
  produced codex's stray top-level tabs.

.mise.toml

@@ -0,0 +1,8 @@
+# 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.
+[tools]
+zig = "0.15.2"

CHANGELOG.md

@@ -6,7 +6,109 @@ loosely follows [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
 ## [Unreleased]
 
+## [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
@@ -64,6 +166,14 @@ loosely follows [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
   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
@@ -71,6 +181,66 @@ loosely follows [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
   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

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)"
 

TODO.md

@@ -1,7 +1,171 @@
-- [ ] We should probably rename the Kill <Process> terminology to Close <Process> instead, across processes and agents.
-- [ ] Exited shells are still being treated as active processes. They should be removed from the process list when they exit.
-- [ ] Shells should be renamed to terminals. "New Terminal" etc.
+# Perf Audit (auto-generated 2026-05-15)
+Findings from a codebase sweep — not user-reported, needs review before
+action. Each item names the anchor and a sketched fix.
 
+Baseline benchmark numbers (`go test -bench=. ./internal/app/`, AMD
+Ryzen 7 7800X3D, libghostty-vt **ReleaseFast** after the Makefile
+fix landed):
+
+```
+# Renderer alone
+ViewportRenderer_PlainASCII       229 MB/s     1.3 KB/op    6 allocs/op
+ViewportRenderer_StyledLines       89 MB/s    91   KB/op  4325 allocs/op
+ViewportRenderer_RatatuiBurst      40 MB/s   365   KB/op 17306 allocs/op
+RendererThroughput_ReuseInstance   90 MB/s   316   KB/op 17380 allocs/op
+ContainsOSC_NoOSC                3050 MB/s     0   B/op     0 allocs/op
+
+# ASCII-video stream (renderer only — 3 sec at the target fps)
+ASCIIVideo_Stream_8Color_120fps     260 µs/frame  3845 fps_ceiling   3.1% budget
+ASCIIVideo_Stream_TrueColor_120fps  576 µs/frame  1735 fps_ceiling   6.9% budget
+
+# Full pipeline (em.Write + renderer + io.Discard write)
+Pipeline_ASCIIVideo_8Color_120fps     493 µs/frame  2030 fps_ceiling   5.9% budget
+Pipeline_ASCIIVideo_TrueColor_120fps 1075 µs/frame   931 fps_ceiling  12.9% budget
+
+# Emulator alone (libghostty-vt CSI/SGR parser)
+Emulator_Write_Stream_8Color_120fps    257 µs/frame  3890 fps_ceiling
+Emulator_Write_Stream_TrueColor_120fps 488 µs/frame  2051 fps_ceiling
+```
+
+Result of the fix below: 27-32× pipeline speedup, 60× emulator
+speedup. Pipeline hits 930-2030 fps end-to-end — 7-16× headroom
+over the 120 fps target on the heaviest workload (truecolor
+full-screen redraws).
+
+
+- [ ] **viewport renderer allocates ~1 alloc per 4 input bytes on SGR/CSI-heavy chunks.** [MEDIUM]
+  - `internal/app/viewport_renderer.go` — the styled-lines and
+    ratatui benchmarks show 4-17k allocs per chunk. The hot
+    contributors are likely (a) `string(vr.buf)` / `string(params)`
+    conversions in `emitCSI` for every escape sequence, (b) the
+    `pending strings.Builder` resizing as fragments arrive, and (c)
+    `vr.shifter.Shift(vr.buf)` returning a fresh slice per CSI.
+  - Fix direction: switch CSI param parsing to byte-slice
+    comparison (no string conversion); reuse `vr.buf` and
+    `vr.pending` backing arrays across `Render` calls by
+    pre-growing in `newViewportRenderer`; have `cursorShifter.Shift`
+    return into a caller-owned buffer instead of allocating.
+    Profile-guided: run the styled-lines bench, point pprof at the
+    allocs profile, fix the top three call sites.
+
+- [ ] **viewport renderer throughput (~90 MB/s styled) limits codex steady-state.** [MEDIUM]
+  - The styled-lines and ratatui benchmarks come in at 89 MB/s and
+    40 MB/s respectively. A 100 KB/s codex burst is far under that
+    limit, but a session-resume dump of a 5 MiB chat history takes
+    50-130 ms of pure renderer time at those rates — enough to be
+    user-visible at the start of a long resume.
+  - Fix direction: same as the alloc fix above; once the per-call
+    allocation cost drops, the throughput ceiling rises with it.
+    Worth re-running the benches after fixing the allocs and only
+    investing further if the styled-lines bench is still under
+    ~300 MB/s.
+
+- [ ] **Session.Children() allocates a fresh slice on every call.** [MEDIUM]
+  - `internal/app/session.go:530-541` walks `s.order` under `s.mu` and
+    builds a new `[]*Child` slice every time. Callers on hot paths:
+    `drawSidebar` calls it twice per frame
+    (`internal/app/sidebar.go:139` and `:171`); `drawTabBar` calls it
+    once per frame (`internal/app/tabbar.go:37`); the classifier
+    iterates it every 250 ms (`internal/app/classifier.go:38`); and
+    palette/navigation hit it on every Ctrl-A/D/W/S keystroke.
+  - Fix direction: store the snapshot in an `atomic.Pointer[[]*Child]`
+    on `Session`, refresh it under `s.mu` only when `Spawn` / `delete`
+    mutates the map. Readers get O(1) `Load()` with zero allocation —
+    same pattern already used for `listeners` (session.go:118-123).
+
+- [ ] **wait_for_pattern re-scans the entire stream/grid every iteration.** [MEDIUM]
+  - `internal/app/host.go:476-493` (the `check` closure). On `scope =
+    "scrollback"` it calls `c.StreamRead(0)` followed by
+    `stripANSIBytes(nil, b)` over the entire ring on every wake — a
+    full O(ring size) walk per chunk arrival. On `grid` it goes
+    through PlainText (one CGO call) plus a regex match against the
+    full grid string. For an agent waiting on a marker in a chatty
+    pane, every PTY chunk fires `check()`.
+  - Fix direction: for `scrollback`, track the offset of the last
+    check and run the regex only over the new tail, reusing a
+    per-call scratch buffer for ANSI stripping. For `grid`, dedupe
+    on `ScreenVersion()` — skip when version hasn't changed.
+
+- [ ] **search_output compiles regex + strips ANSI on every call.** [MEDIUM]
+  - `internal/app/host.go:428` compiles a fresh `regexp.Regexp` per
+    invocation; `:434` strips ANSI over the entire ring buffer when
+    `kind="rendered"`. Agents that poll `search_output` with the same
+    pattern (the typical "watch for marker" loop) repay both costs on
+    every call.
+  - Fix direction: small LRU of compiled regexes keyed by pattern
+    string (cap maybe 32) on `toolHost`. Cache the stripped-ANSI
+    buffer keyed by `c.ScreenVersion()` so consecutive searches over
+    an unchanged ring reuse the strip.
+
+- [ ] **GetProcessOutput grid mode acquires the emulator twice.** [MEDIUM]
+  - `internal/app/host.go:375-391` does `em := c.Emulator()` for
+    ActiveScreen / Cursor / Size, then at line 387 re-fetches
+    `em := c.Emulator()` for PlainText. Each `Emulator()` call goes
+    through `ptyMu` and inspects the live PTY pointer. Under a
+    chatty agent polling `get_process_output` every 100 ms this is
+    a redundant lock and pointer chase per call.
+  - Fix direction: hold the emulator reference from the first
+    lookup; reuse it for PlainText. The check `if em == nil` still
+    runs cleanly because the variable is captured.
+
+- [ ] **FindChildByIdentity is O(N) under the session lock.** [LOW]
+  - `internal/app/session.go:553-565` scans the children map looking
+    for a matching `Identity` token on every new mcp-stdio
+    connection. Not a steady-state hot path — only fires once per
+    child spawn — but with many short-lived sub-agents it adds up
+    and contends with everyone else taking `s.mu`.
+  - Fix direction: maintain an `identityIndex map[string]string`
+    (identity → child id) updated alongside spawn / exit, give the
+    lookup an O(1) read.
+
+- [ ] **Per-promoter regex matches in the idle classifier.** [LOW]
+  - `internal/app/idle.go:175-182` (`matchAny`) walks each compiled
+    pattern and runs the DFA over the same 4 KiB tail. A preset with
+    five permission patterns + five error patterns is ten DFA
+    invocations per child per 250 ms tick.
+  - Fix direction: at preset load time, compile each `_patterns`
+    list into a single alternation regex (`(?:p1)|(?:p2)|…`). The
+    classifier then makes one Match call per category per tick.
+
+- [ ] **Port-detection dedup is O(N²) over c.ports.** [LOW]
+  - `internal/app/child.go:461-467`: for each fresh URL match the
+    code linearly scans the existing port list. The list rarely
+    grows past a handful, but a dev server that lists "all open
+    ports" in one log line interacts badly: M new matches × N
+    existing entries.
+  - Fix direction: keep a `seenPorts map[int]struct{}` next to
+    `c.ports`, rebuilt on prune (none today). O(1) per match.
+
+- [ ] **Port-sighting string allocations happen before the dedup check.** [LOW]
+  - `internal/app/child.go:455-456` allocates `urlForm` and `portStr`
+    before line 461's `seen` walk. Both strings are wasted when the
+    port is already in `c.ports`. Inside `c.portsMu` for the whole
+    loop body too, blocking the `Ports()` reader path.
+  - Fix direction: bind the port int first (cheap parse from
+    `m[1]`), do the seen check, only then allocate the URL string
+    for the surviving sighting.
+
+- [ ] **classifier `time.Now()` syscall per child per tick.** [LOW]
+  - `internal/app/classifier.go:54` (and the `IdleMS` /
+    `TitleIdleMS` helpers it transitively calls in
+    `internal/app/child.go:343-374`) each call `time.Now()`.
+    Reading time on Linux is fast (vDSO) but with N children × 4
+    `time.Now()` per tick × 4 ticks/sec it's wasted work that can
+    be batched.
+  - Fix direction: capture `now := time.Now().UnixNano()` once at
+    the top of `classifyAll` and thread it into `classifyOne` and
+    the helpers as a parameter.
+
+- [ ] **wait_for_pattern subscribes a listener for every call.** [LOW]
+  - `internal/app/host.go:472-474`: each invocation calls
+    `Session.Subscribe(wake)` which clones the listener slice and
+    swaps the atomic pointer; the `defer Unsubscribe` does the same
+    on exit. Two allocations per `wait_for_pattern`. The agent
+    pattern of looping on `wait_for_pattern` after every tool call
+    pays this churn on the steady-state path.
+  - Fix direction: a per-child `chunkBroadcaster` registered once
+    at child spawn that hands out lightweight subscriber tokens,
+    rather than going through the full session listener machinery.
 
 # On Hold
 - [ ] There's a unicode <?> being displayed in opencode [ON HOLD]

cmd/patterm/main.go

@@ -16,7 +16,10 @@ import (
 	"context"
 	"fmt"
 	"os"
+	"path/filepath"
+	"runtime"
 	"runtime/debug"
+	"runtime/pprof"
 	"time"
 
 	flag "github.com/spf13/pflag"
@@ -49,7 +52,13 @@ func main() {
 	var (
 		projectDir  = flag.String("project", "", "project directory (default $PWD)")
 		showVersion = flag.Bool("version", false, "print version and exit")
+		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 {
@@ -73,15 +82,104 @@ 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{
 		ProjectDir: cwd,
 		ProjectKey: key,
+		DebugDir:   resolvedDebug,
+		ProfileDir: resolvedProfile,
 	}); 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")

internal/app/app.go

@@ -29,6 +29,17 @@ import (
 type Options struct {
 	ProjectDir string
 	ProjectKey string
+	// DebugDir, when non-empty, enables verbose debug logging to
+	// <DebugDir>/patterm.log and per-child raw PTY output capture to
+	// <DebugDir>/<child-id>.raw. The dir is created if missing. Events
+	// (spawn / exit / state change) land in <DebugDir>/events.jsonl.
+	DebugDir string
+	// ProfileDir, when non-empty, enables in-process performance
+	// counters. patterm writes a per-second JSONL snapshot stream to
+	// <ProfileDir>/metrics.jsonl, a final aggregate to metrics.json,
+	// and a human-readable summary.txt on shutdown. The pprof files
+	// written by --profile sit alongside these in the same dir.
+	ProfileDir string
 }
 
 const keyCtrlK byte = 0x0b
@@ -77,6 +88,22 @@ func Run(ctx context.Context, opts Options) error {
 
 	sess := NewSession(opts.ProjectDir, opts.ProjectKey)
 	defer sess.Shutdown()
+
+	// Debug capture: when --debug=<dir> is set, write a verbose log
+	// (patterm.log), per-child raw PTY output (<id>.raw), and a
+	// JSONL event stream (events.jsonl). Installed before the TUI
+	// listener so the very first OnChildSpawned / OnPTYOut event
+	// is captured.
+	if opts.DebugDir != "" {
+		dc, err := openDebugCapture(opts.DebugDir)
+		if err != nil {
+			return fmt.Errorf("app: debug capture: %w", err)
+		}
+		os.Setenv("PATTERM_DEBUG_LOG", dc.LogPath())
+		sess.Subscribe(dc)
+		defer dc.Close()
+		logf("debug capture enabled at %s", opts.DebugDir)
+	}
 	// Snapshot persisted processes BEFORE attaching the store: Spawn
 	// mints fresh ids, so the old records would otherwise linger
 	// alongside the new ones. Drop them up front; the restore loop
@@ -113,6 +140,23 @@ func Run(ctx context.Context, opts Options) error {
 	ctx, cancel := context.WithCancel(ctx)
 	defer cancel()
 
+	// Performance tracker — instrumented hot-path timings written to
+	// <ProfileDir>. nil when --profile is off, in which case every
+	// record*() call is a fast nil check.
+	metrics, err := newMetricsTracker(opts.ProfileDir)
+	if err != nil {
+		return fmt.Errorf("app: metrics tracker: %w", err)
+	}
+	if metrics != nil {
+		go metrics.run(ctx)
+		defer metrics.close()
+	}
+
+	// Per-session idle-detection classifier. One goroutine ticks every
+	// 250ms over every live child and updates IdleState. It stops when
+	// ctx is cancelled.
+	go sess.runClassifier(ctx)
+
 	st := &uiState{
 		sess:       sess,
 		presets:    presets,
@@ -120,10 +164,13 @@ func Run(ctx context.Context, opts Options) error {
 		pads:       pads,
 		chromeWake: make(chan struct{}, 1),
 		trust:      trustStore,
+		timers:     host.timers,
 		hostCols:   cols,
 		hostRows:   rows,
 		stdinTTY:   term.IsTerminal(int(os.Stdin.Fd())),
+		metrics:    metrics,
 	}
+	sess.SetMetrics(metrics)
 	host.attention = st
 	host.focus = st
 	host.prompter = st
@@ -242,11 +289,20 @@ func Run(ctx context.Context, opts Options) error {
 			case <-st.chromeWake:
 			case <-ticker.C:
 			}
-			if !st.chromeDirty.Swap(false) {
+			chromeChanged := st.chromeDirty.Swap(false)
+			sidebarChanged := st.sidebarDirty.Swap(false)
+			didWork := chromeChanged || sidebarChanged
+			st.metrics.recordTickerFire(didWork)
+			if !didWork {
 				continue
 			}
-			st.drawTabBar()
-			st.drawStatusLine()
+			if chromeChanged {
+				st.drawTabBar()
+				st.drawStatusLine()
+			}
+			if sidebarChanged {
+				st.drawSidebar()
+			}
 		}
 	}()
 
@@ -296,6 +352,7 @@ type uiState struct {
 	launcher *Launcher
 	pads     *scratchpad.Store
 	trust    *trust.Store
+	timers   *timerManager
 
 	outMu sync.Mutex
 
@@ -348,6 +405,11 @@ type uiState struct {
 	hostCols, hostRows uint16
 	stdinTTY           bool
 
+	// metrics is the optional performance tracker. nil when --profile
+	// is off. Hot paths call metrics.recordX which is a fast nil
+	// check on the disabled path.
+	metrics *metricsTracker
+
 	// chromeCacheMu guards the last-rendered byte cache for each chrome
 	// element. The tab bar, sidebar, and status line all repaint on
 	// many state changes and on every PTY chunk, but their content
@@ -365,7 +427,14 @@ type uiState struct {
 	// sensitive paths (owner flip, attention, trust, focus change)
 	// continue to call drawStatusLine / drawTabBar synchronously.
 	chromeDirty atomic.Bool
-	chromeWake  chan struct{}
+	// sidebarDirty defers sidebar repaints off the per-chunk hot path
+	// in the same way. A long claude session resume — where every PTY
+	// chunk scrolls the viewport — used to call drawSidebar()
+	// synchronously per chunk, which dominated the resume's wall time
+	// (hundreds of full-sidebar rebuilds for a frame that was almost
+	// always cache-equal).
+	sidebarDirty atomic.Bool
+	chromeWake   chan struct{}
 
 	// padsCacheMu guards the cached scratchpad listing. The sidebar
 	// and palette/sidebar nav helpers read it on every chunk-driven
@@ -408,14 +477,18 @@ func (st *uiState) focusProcess(processID string) {
 		return
 	}
 	layout := st.layoutSnapshot()
+	onAlt := childIsOnAlt(c)
 	st.mu.Lock()
 	leavingPad := st.focusedPad != ""
 	st.focusedPad = ""
 	st.focusedID = c.ID
 	st.focusedName = c.DisplayName()
 	st.updateActiveAgentLocked(c)
-	st.renderer = newViewportRenderer(layout)
+	r := newViewportRenderer(layout)
+	r.SetChildOnAlt(onAlt)
+	st.renderer = r
 	st.mu.Unlock()
+	st.syncHostMouseForChild(onAlt)
 	// Wipe whatever the previous focus (PTY child or pad view) left in
 	// the viewport before painting the new child's snapshot.
 	if leavingPad {
@@ -427,6 +500,41 @@ func (st *uiState) focusProcess(processID string) {
 	st.drawStatusLine()
 }
 
+// childIsOnAlt reports whether the child's emulator is currently on
+// its alternate screen. Returns false if the emulator is gone or the
+// query fails.
+func childIsOnAlt(c *Child) bool {
+	if c == nil {
+		return false
+	}
+	em := c.Emulator()
+	if em == nil {
+		return false
+	}
+	sc, err := em.ActiveScreen()
+	if err != nil {
+		return false
+	}
+	return sc == vt.ScreenAlternate
+}
+
+// syncHostMouseForChild emits the host mouse-reporting toggle that
+// matches a newly-focused child's screen side. Primary-screen children
+// want host mouse armed so the wheel drives inline scrollback; alt-
+// screen children get host mouse disabled by default so click-and-drag
+// selection works. Alt-screen TUIs that need mouse (vim, ranger, etc.)
+// re-enable it themselves, and the viewport renderer forwards those
+// toggles back to the host.
+func (st *uiState) syncHostMouseForChild(onAlt bool) {
+	st.outMu.Lock()
+	defer st.outMu.Unlock()
+	if onAlt {
+		_, _ = os.Stdout.WriteString("\x1b[?1000l\x1b[?1006l")
+	} else {
+		_, _ = os.Stdout.WriteString("\x1b[?1000h\x1b[?1006h")
+	}
+}
+
 // focusScratchpad shifts focus to a scratchpad. The main viewport
 // renders the pad's text instead of any child PTY; PTY output for the
 // previously focused child is dropped until focus moves back to a
@@ -565,12 +673,14 @@ func (st *uiState) scratchpadsChanged() {
 // OnChildSpawned auto-focuses the new child.
 func (st *uiState) OnChildSpawned(c *Child) {
 	layout := st.layoutSnapshot()
+	onAlt := childIsOnAlt(c)
 	st.mu.Lock()
 	st.focusedPad = ""
 	st.focusedID = c.ID
 	st.focusedName = c.DisplayName()
 	st.updateActiveAgentLocked(c)
 	renderer := newViewportRenderer(layout)
+	renderer.SetChildOnAlt(onAlt)
 	st.renderer = renderer
 	palOpen := st.palette != nil
 	if palOpen {
@@ -604,12 +714,21 @@ func (st *uiState) OnChildSpawned(c *Child) {
 		st.outMu.Unlock()
 	}
 
+	st.syncHostMouseForChild(onAlt)
 	st.moveToViewportOrigin()
 	st.drawTabBar()
 	st.drawSidebar()
 	st.drawStatusLine()
 }
 
+// OnChildStateChanged repaints the sidebar whenever a child's
+// idle-state badge flips. Cheap — the badge is the only chrome that
+// reflects state today, and drawSidebar bails when the cached frame
+// hasn't changed.
+func (st *uiState) OnChildStateChanged(string, IdleState) {
+	st.drawSidebar()
+}
+
 // OnChildExited drops focus and shows the empty state if it was the
 // focused child.
 func (st *uiState) OnChildExited(c *Child) {
@@ -695,6 +814,10 @@ func (st *uiState) scheduleAutoRestart(c *Child) {
 // disabled only around the replay so long styled runs cannot wrap into
 // the right rail.
 func (st *uiState) OnPTYOut(childID string, chunk []byte) {
+	var entry time.Time
+	if st.metrics != nil {
+		entry = time.Now()
+	}
 	layout := st.layoutSnapshot()
 	st.mu.Lock()
 	focus := st.focusedID
@@ -711,16 +834,31 @@ func (st *uiState) OnPTYOut(childID string, chunk []byte) {
 	}
 	st.mu.Unlock()
 	if palOpen || focus != childID || renderer == nil {
+		st.metrics.recordPTYOutDrop()
 		return
 	}
 	var out []byte
 	if forceRepaint {
+		var snapStart time.Time
+		if st.metrics != nil {
+			snapStart = time.Now()
+		}
 		out = st.renderFocusedSnapshot(childID, renderer, layout)
+		if st.metrics != nil {
+			st.metrics.recordSnapshot(time.Since(snapStart))
+		}
 		if len(out) == 0 {
 			return
 		}
 	} else {
+		var rstart time.Time
+		if st.metrics != nil {
+			rstart = time.Now()
+		}
 		out = renderer.Render(chunk)
+		if st.metrics != nil {
+			st.metrics.recordRender(time.Since(rstart))
+		}
 	}
 	// One write covers the autowrap-disable prelude, the chunk, and the
 	// autowrap-restore postlude — three syscalls collapsed into one
@@ -730,9 +868,16 @@ func (st *uiState) OnPTYOut(childID string, chunk []byte) {
 	wrapped = append(wrapped, "\x1b[?7l"...)
 	wrapped = append(wrapped, out...)
 	wrapped = append(wrapped, "\x1b[?7h"...)
+	var wstart time.Time
+	if st.metrics != nil {
+		wstart = time.Now()
+	}
 	st.outMu.Lock()
 	_, _ = os.Stdout.Write(wrapped)
 	st.outMu.Unlock()
+	if st.metrics != nil {
+		st.metrics.recordStdout(time.Since(wstart), len(wrapped))
+	}
 	// RI / IND / NEL / SU / SD / IL / DL and bottom-margin LF / VT / FF
 	// scroll content within the host's scroll region, which spans every
 	// column — so any of them drags the right-hand sidebar's session-tree
@@ -745,15 +890,23 @@ func (st *uiState) OnPTYOut(childID string, chunk []byte) {
 		st.chromeCacheMu.Lock()
 		st.sidebarCache = ""
 		st.chromeCacheMu.Unlock()
-		// Scrolled chunks can clobber the sidebar columns; repaint
-		// synchronously so the gap fills before the next chunk lands.
-		st.drawSidebar()
+		// Defer the sidebar repaint to the chrome ticker. On a long
+		// session resume every PTY chunk scrolls, and a synchronous
+		// drawSidebar() per chunk dominates wall time even when the
+		// frame ends up cache-equal — the rebuild work is unconditional.
+		// The chrome ticker drains the dirty flag at ~60 Hz, so the
+		// visible gap a scrolled chunk can leave in the sidebar columns
+		// is bounded by one frame.
+		st.markSidebarDirty()
 	}
 	// Defer the tab bar + status line repaint to the chrome ticker.
 	// The cached frame already short-circuits the wire write, but
 	// avoiding the string build, FindChild, and locking on every
 	// chunk pulls steady-state CPU off the hot path.
 	st.markChromeDirty()
+	if st.metrics != nil {
+		st.metrics.recordPTYOut(time.Since(entry), len(chunk))
+	}
 }
 
 func (st *uiState) enterScreen() {
@@ -851,6 +1004,18 @@ func (st *uiState) markChromeDirty() {
 	}
 }
 
+// markSidebarDirty schedules a sidebar repaint on the next ticker
+// frame. Hot path — every scrolled PTY chunk lands here. Synchronous
+// repaints from latency-sensitive sites (spawn, exit, focus, state
+// change, trust) keep calling drawSidebar directly.
+func (st *uiState) markSidebarDirty() {
+	st.sidebarDirty.Store(true)
+	select {
+	case st.chromeWake <- struct{}{}:
+	default:
+	}
+}
+
 func (st *uiState) invalidateChromeCache() {
 	st.chromeCacheMu.Lock()
 	st.tabBarCache = ""
@@ -881,6 +1046,10 @@ func (st *uiState) renderPaletteLocked() {
 // attention ask. Right side: palette hint. The PTY child occupies
 // host_rows-1 rows so this row is exclusively ours.
 func (st *uiState) drawStatusLine() {
+	var entry time.Time
+	if st.metrics != nil {
+		entry = time.Now()
+	}
 	st.mu.Lock()
 	palOpen := st.palette != nil
 	focusID := st.focusedID
@@ -967,10 +1136,16 @@ func (st *uiState) drawStatusLine() {
 	st.chromeCacheMu.Lock()
 	if line == st.statusLineCache {
 		st.chromeCacheMu.Unlock()
+		if st.metrics != nil {
+			st.metrics.recordStatus(time.Since(entry), true)
+		}
 		return
 	}
 	st.statusLineCache = line
 	st.chromeCacheMu.Unlock()
+	if st.metrics != nil {
+		defer func() { st.metrics.recordStatus(time.Since(entry), false) }()
+	}
 
 	st.outMu.Lock()
 	defer st.outMu.Unlock()
@@ -1607,6 +1782,13 @@ func (st *uiState) closePalette(action paletteAction) {
 			st.flashError(fmt.Sprintf("spawn %s: %v", action.preset.Name, err))
 		}
 
+	case "spawn-terminal":
+		l := st.layoutSnapshot()
+		st.launcher.SetSize(l.childCols(), l.childRows())
+		if _, err := st.launcher.LaunchTerminal(nil, "terminal", "", "", nil); err != nil {
+			st.flashError(fmt.Sprintf("spawn terminal: %v", err))
+		}
+
 	case "spawn-process-submit":
 		if action.command == "" {
 			restoreView()

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

@@ -123,6 +123,19 @@ type Child struct {
 	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
@@ -330,6 +343,75 @@ 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)

internal/app/classifier.go

@@ -0,0 +1,96 @@
+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 := c.tailBytes(classifierTailBytes)
+	state, reason := classify(c.idleDetection, exited, exitNonZero, idleMS, titleIdleMS, title, tail)
+	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/debug.go

@@ -0,0 +1,155 @@
+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) 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

@@ -61,12 +61,11 @@ 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 +75,28 @@ 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 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 (h *toolHost) SetSize(cols, rows uint16) {
@@ -531,6 +552,7 @@ func (n *chunkNotifier) OnPTYOut(id string, chunk []byte) {
 	default:
 	}
 }
+func (n *chunkNotifier) OnChildStateChanged(string, IdleState) {}
 
 func (h *toolHost) GetProcessPorts(callerID, processID string) ([]mcp.PortSighting, error) {
 	c := h.sess.FindChild(processID)
@@ -725,27 +747,59 @@ 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
 }
 
 // ───────────────────────────────────────────────────────────────────
@@ -816,6 +870,10 @@ 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
 }
 
@@ -1026,7 +1084,9 @@ 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",
+		"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",
 		"whoami", "help",
 	}
@@ -1051,13 +1111,13 @@ 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":
@@ -1086,9 +1146,18 @@ func helpFor(topic string) mcp.HelpResponse {
 		}
 	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{

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 +1,225 @@
+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
+func classify(cfg *resolvedIdleDetection, exited, exitNonZero bool, idleMS, titleIdleMS int64, title string, tail []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 {
+		if matchAny(cfg.errorRegexes, tail) {
+			return StateError, "error regex matched"
+		}
+		if matchAny(cfg.permissionRegexes, tail) {
+			return StatePermission, "permission regex matched"
+		}
+		if matchAny(cfg.thinkingRegexes, tail) {
+			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, tail []byte) bool {
+	for _, re := range res {
+		if re.Match(tail) {
+			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 +1,112 @@
+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)
+			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); 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); 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); got != StateWorking {
+		t.Fatalf("no title yet, recent output: got %q", got)
+	}
+	if got, _ := classify(cfg, false, false, 5000, 0, "", nil); got != StateIdle {
+		t.Fatalf("no title yet, output idle: got %q", got)
+	}
+}
+
+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); got != StateThinking {
+		t.Fatalf("thinking title: got %q", got)
+	}
+	if got, _ := classify(cfg, false, false, 9999, 500, "Waiting for permission", 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); 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]")); got != StatePermission {
+		t.Fatalf("permission promoter: got %q", got)
+	}
+	// Error trumps permission.
+	if got, _ := classify(cfg, false, false, 5000, 0, "", []byte("panic: bad\nApprove?")); 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…")); got != StateThinking {
+		t.Fatalf("thinking promoter: got %q", got)
+	}
+}
+
+func TestClassifyExitTerminal(t *testing.T) {
+	cfg := &resolvedIdleDetection{strategy: StrategyOutputActivity, idleThresholdMS: 2000}
+	if got, _ := classify(cfg, true, true, 0, 0, "", nil); got != StateError {
+		t.Fatalf("non-zero exit: got %q", got)
+	}
+	if got, _ := classify(cfg, true, false, 0, 0, "", nil); got != StateIdle {
+		t.Fatalf("clean exit: got %q", got)
+	}
+}

internal/app/launch.go

@@ -127,14 +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()
@@ -171,15 +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

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.go

@@ -11,12 +11,13 @@ import (
 // paletteAction is what the palette returns when the user picks an item.
 type paletteAction struct {
 	// kind: "spawn-agent" | "spawn-process" | "spawn-process-form" |
-	//       "spawn-process-submit" | "switch" | "kill" | "quit" |
-	//       "cancel" | "pad-delete" | "pad-rename" | "pad-rename-form" |
-	//       "pad-rename-submit" | "pad-edit" | "agent-rename" |
-	//       "agent-rename-form" | "agent-rename-submit" | "agent-close" |
-	//       "proc-rename" | "proc-rename-form" | "proc-rename-submit" |
-	//       "proc-delete" | "proc-stop" | "proc-restart"
+	//       "spawn-process-submit" | "spawn-terminal" | "switch" |
+	//       "kill" | "quit" | "cancel" | "pad-delete" | "pad-rename" |
+	//       "pad-rename-form" | "pad-rename-submit" | "pad-edit" |
+	//       "agent-rename" | "agent-rename-form" | "agent-rename-submit" |
+	//       "agent-close" | "proc-rename" | "proc-rename-form" |
+	//       "proc-rename-submit" | "proc-delete" | "proc-stop" |
+	//       "proc-restart"
 	kind string
 
 	// For spawn-agent / spawn-process, the preset to launch.
@@ -276,6 +277,16 @@ func (p *paletteState) allItems() []paletteItem {
 		})
 	}
 
+	// "New Terminal" — bare interactive $SHELL pane. Distinct from
+	// "Run process: …" presets in that it spawns a KindTerminal (which
+	// disappears from the sidebar on exit rather than sticking around
+	// for restart). One quick keystroke; no form.
+	out = append(out, paletteItem{
+		label:  "New Terminal",
+		hint:   "bare interactive $SHELL · removed on exit",
+		action: paletteAction{kind: "spawn-terminal"},
+	})
+
 	// Freeform "Spawn process…" entry. Opens a sub-form for typing an
 	// arbitrary command line and ticking "relaunch on exit". The action
 	// kind is intercepted by acceptOrEnterForm so accept switches the
@@ -288,14 +299,14 @@ func (p *paletteState) allItems() []paletteItem {
 		action: paletteAction{kind: "spawn-process-form"},
 	})
 
-	// Kill entries last among the action rows, before Quit. Mirror the
+	// Close entries last among the action rows, before Quit. Mirror the
 	// "(current)" marker from switch entries so the focused tab is
-	// obvious when scanning the kill list.
+	// obvious when scanning the close list.
 	for _, c := range p.children {
 		if c.Status() != StatusRunning {
 			continue
 		}
-		label := "Kill " + c.DisplayName()
+		label := "Close " + c.DisplayName()
 		if c.ID == p.focused {
 			label = "• " + label + " (current)"
 		}

internal/app/session.go

@@ -50,6 +50,11 @@ type Session struct {
 	// 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 /
@@ -61,6 +66,18 @@ func (s *Session) SetPersistStore(p *persist.Store) {
 	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
 // events without polling.
 type ChildEventListener interface {
@@ -70,6 +87,10 @@ 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)
 }
 
 func NewSession(projectDir, projectKey string) *Session {
@@ -140,6 +161,12 @@ func (s *Session) emitPTYOut(id string, chunk []byte) {
 	}
 }
 
+func (s *Session) emitStateChanged(id string, state IdleState) {
+	for _, l := range s.listenersSnapshot() {
+		l.OnChildStateChanged(id, state)
+	}
+}
+
 func (s *Session) ChildEnv() []string {
 	env := os.Environ()
 	// Mark patterm-owned PTYs so a recursive `patterm` invocation can
@@ -168,6 +195,11 @@ 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
@@ -198,6 +230,12 @@ 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()
@@ -371,9 +409,38 @@ func (s *Session) pumpChild(c *Child, runID uint64) {
 			}
 			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)
@@ -403,6 +470,23 @@ 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()
+	}
 }
 
 // killDescendantsOf terminates every still-live direct child of
@@ -632,6 +716,24 @@ 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

@@ -57,6 +57,50 @@ 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 waitUntilLive(t *testing.T, c *Child) {
 	t.Helper()
 	deadline := time.Now().Add(5 * time.Second)

internal/app/sidebar.go

@@ -4,6 +4,7 @@ import (
 	"fmt"
 	"os"
 	"strings"
+	"time"
 )
 
 const (
@@ -11,6 +12,24 @@ const (
 	statusRows  = 1
 )
 
+// 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.
@@ -19,6 +38,10 @@ 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
@@ -62,14 +85,56 @@ 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,
@@ -92,9 +157,9 @@ func (st *uiState) drawSidebar() {
 		var line string
 		if focused {
 			line = " " + styleAccent + "▎" + styleReset + " " + glyph + " " +
-				styleBold + c.DisplayName() + styleReset + marker
+				styleBold + c.DisplayName() + styleReset + marker + timerIndicator(c)
 		} else {
-			line = "   " + glyph + " " + styleHint + c.DisplayName() + styleReset + marker
+			line = "   " + glyph + " " + styleHint + c.DisplayName() + styleReset + marker + timerIndicator(c)
 		}
 		write(line)
 	}
@@ -124,9 +189,9 @@ func (st *uiState) drawSidebar() {
 		var line string
 		if focused {
 			line = " " + styleAccent + "▎" + styleReset + " " + indent + glyph + " " +
-				styleBold + c.DisplayName() + styleReset
+				styleBold + c.DisplayName() + styleReset + timerIndicator(c)
 		} else {
-			line = "   " + indent + glyph + " " + styleHint + c.DisplayName() + styleReset
+			line = "   " + indent + glyph + " " + styleHint + c.DisplayName() + styleReset + timerIndicator(c)
 		}
 		write(line)
 	}
@@ -170,10 +235,16 @@ 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.

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/tabbar.go

@@ -4,6 +4,7 @@ import (
 	"fmt"
 	"os"
 	"strings"
+	"time"
 	"unicode/utf8"
 )
 
@@ -17,6 +18,10 @@ 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
@@ -188,10 +193,16 @@ func (st *uiState) drawTabBar() {
 	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()

internal/app/timers.go

@@ -0,0 +1,542 @@
+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
+
+	// 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),
+	}
+	m.fireFn = defaultFireFn
+	return m
+}
+
+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) })
+	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.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()
+	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.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()
+	for _, f := range fires {
+		m.fireFn(f.owner, f.body, f.label)
+	}
+}
+
+// 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()
+	defer m.mu.Unlock()
+	t, ok := m.timers[id]
+	if !ok {
+		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 {
+		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.
+		return nil
+	}
+	if t.rt != nil {
+		t.rt.Stop()
+		t.rt = nil
+	}
+	t.status = timerStatusCanceled
+	delete(m.timers, id)
+	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()
+	defer m.mu.Unlock()
+	t, ok := m.timers[id]
+	if !ok {
+		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 {
+		return mcp.Errorf(mcp.ErrorKindRoleForbidden, "timer %q is not owned by caller", id)
+	}
+	if t.status != timerStatusPending {
+		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
+	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()
+	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
+}
+
+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 +1,413 @@
+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 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)
+	}
+}

internal/app/tree.go

@@ -96,17 +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

internal/app/viewport_renderer.go

@@ -33,6 +33,14 @@ type viewportRenderer struct {
 	// 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
@@ -65,6 +73,16 @@ func newViewportRenderer(l terminalLayout) *viewportRenderer {
 	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) {
 	vr.mu.Lock()
 	defer vr.mu.Unlock()
@@ -236,15 +254,36 @@ func (vr *viewportRenderer) emitCSI() {
 			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) {
-			// Patterm owns mouse reporting on the host so wheel events keep
-			// flowing for scroll-viewport. The child's own emulator still
-			// observes the mode set/reset (it processes the same bytes we
-			// hand to ghostty_terminal_vt_write), so we know whether the
-			// child wants mouse input — we just don't let it disarm our
-			// host listener.
+			// 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
 		}
 	}

internal/app/viewport_renderer_test.go

@@ -24,8 +24,36 @@ 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)
 	}
 }
 

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/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/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/mcp/protocol.go

@@ -27,6 +27,24 @@ var serverInfo = map[string]any{
 	"version": "0.1.0",
 }
 
+// serverInstructions is returned in the MCP `initialize` response. MCP
+// clients show this to the underlying LLM as context for how to use
+// the server. Failure modes we've seen and want to head off:
+//   - The agent assumes patterm is something it has to launch (running
+//     `patterm` or `patterm mcp-stdio` from its own shell). It's
+//     already attached — it just calls the tools.
+//   - The agent reaches for shell tools (perl / nc / socat / curl) to
+//     poke patterm's Unix socket directly. That socket connection
+//     carries no caller identity, so any sub-agent the agent spawns
+//     that way ends up as a stray top-level tab instead of a child
+//     under the spawning agent. Always go through the MCP tools.
+//   - The agent shells out to `claude` / `codex` / `opencode` to start
+//     a peer instead of calling `spawn_agent`. Those peers won't show
+//     up as sub-agents and won't be tied into the patterm lifecycle.
+//
+// Keep this short — clients vary in how much they surface to the LLM.
+const serverInstructions = "You are already running INSIDE patterm; the `patterm` MCP server is connected over the same stdio MCP transport you use for any other MCP server. Use the MCP tools you see in tools/list — do NOT (a) try to launch `patterm` or `patterm mcp-stdio` yourself, (b) poke the Unix socket through perl / nc / socat / curl, or (c) shell out to `claude` / `codex` / `opencode` to start a peer. Any of those bypasses caller-identity and the new agent will land as a stray top-level tab instead of a child under you. Start with `whoami` for your role and the full tool list, then `help('topics')` for orientation. `spawn_agent` is the only correct way to start a sub-agent; `spawn_process` is for non-LLM commands; `list_processes` / `get_process_output` inspect them; `send_input` / `send_message` drive them. Whatever you spawn is yours to `close_process` when done."
+
 // toolDescriptor is the shape returned by `tools/list`. inputSchema is
 // a JSON Schema object — we provide a minimal `{type: "object"}` schema
 // for each tool, which lets MCP clients accept arbitrary arguments and
@@ -73,6 +91,14 @@ func booleanProp(desc string) map[string]any {
 	return map[string]any{"type": "boolean", "description": desc}
 }
 
+func arrayOfStringsProp(desc string) map[string]any {
+	return map[string]any{
+		"type":        "array",
+		"description": desc,
+		"items":       map[string]any{"type": "string"},
+	}
+}
+
 // toolCatalog is the full list advertised via tools/list. Descriptions
 // are intentionally short — clients are expected to fetch help() for
 // detail. Schemas mirror the param structs in tools.go.
@@ -80,7 +106,7 @@ func toolCatalog() []toolDescriptor {
 	return []toolDescriptor{
 		{
 			Name:        "spawn_agent",
-			Description: "Spawn a sub-agent from an agent preset and optionally seed it with initial instructions. Caller owns lifecycle: when the sub-agent's work is done (it reports back via send_message, or you no longer need it), call close_process on its process_id to free the pane and tear down the PTY. See help('lifecycle').",
+			Description: "Spawn a sub-agent from an agent preset and optionally seed it with initial instructions. This is the ONLY correct way to start a sub-agent under you — do not shell out to `claude` / `codex` / `opencode` and do not poke patterm's Unix socket via perl / nc / socat. Either bypasses caller identity and the new agent lands as a stray top-level tab instead of your child. Caller owns lifecycle: when the sub-agent's work is done (it reports back via send_message, or you no longer need it), call close_process on its process_id to free the pane and tear down the PTY. See help('spawning') and help('lifecycle').",
 			InputSchema: objectSchema(map[string]any{
 				"agent":              stringProp("Preset name (e.g. \"claude\", \"codex\")."),
 				"agent_instructions": stringProp("Initial prompt typed into the agent after it's ready."),
@@ -239,12 +265,70 @@ func toolCatalog() []toolDescriptor {
 		},
 		{
 			Name:        "timer_wait",
-			Description: "Sleep server-side for `seconds` and return a timer id (use to pace polling).",
+			Description: "Schedule a delay timer that injects a fixed `[system]` line into your pane when it fires (legacy; prefer timer_set).",
 			InputSchema: objectSchema(map[string]any{
-				"seconds": numberProp("Sleep duration."),
+				"seconds": numberProp("Delay duration."),
 				"label":   stringProp("Optional label for diagnostics."),
 			}, []string{"seconds"}),
 		},
+		{
+			Name:        "timer_set",
+			Description: "Schedule a one-shot delay timer that delivers `body` to the owning agent as a fresh user turn when it fires.",
+			InputSchema: objectSchema(map[string]any{
+				"seconds":          numberProp("Delay duration."),
+				"body":             stringProp("Message delivered verbatim to the owning agent as a user turn when the timer fires."),
+				"label":            stringProp("Optional label for diagnostics."),
+				"owner_process_id": stringProp("Owner process id; defaults to the caller. Top-level callers must supply this explicitly."),
+			}, []string{"seconds", "body"}),
+		},
+		{
+			Name:        "timer_fire_when_idle_any",
+			Description: "Schedule a timer that fires when any watched process enters idle (already-idle entries excluded), or when max_wait_seconds elapses.",
+			InputSchema: objectSchema(map[string]any{
+				"watched":          arrayOfStringsProp("Process ids to watch."),
+				"body":             stringProp("Message delivered verbatim to the owning agent when the timer fires."),
+				"label":            stringProp("Optional label for diagnostics."),
+				"max_wait_seconds": numberProp("Optional cap; 0 means no fallback fire."),
+				"owner_process_id": stringProp("Owner process id; defaults to the caller."),
+			}, []string{"watched", "body"}),
+		},
+		{
+			Name:        "timer_fire_when_idle_all",
+			Description: "Schedule a timer that fires when all watched processes are idle (already-idle entries count as satisfied), or when max_wait_seconds elapses.",
+			InputSchema: objectSchema(map[string]any{
+				"watched":          arrayOfStringsProp("Process ids to watch."),
+				"body":             stringProp("Message delivered verbatim to the owning agent when the timer fires."),
+				"label":            stringProp("Optional label for diagnostics."),
+				"max_wait_seconds": numberProp("Optional cap; 0 means no fallback fire."),
+				"owner_process_id": stringProp("Owner process id; defaults to the caller."),
+			}, []string{"watched", "body"}),
+		},
+		{
+			Name:        "timer_cancel",
+			Description: "Cancel one pending timer owned by the caller.",
+			InputSchema: objectSchema(map[string]any{
+				"timer_id": stringProp("Timer id returned by a previous timer_* call."),
+			}, []string{"timer_id"}),
+		},
+		{
+			Name:        "timer_pause",
+			Description: "Pause one pending timer owned by the caller. Idle-aware timers stop listening to state changes; delay timers preserve their remaining wait.",
+			InputSchema: objectSchema(map[string]any{
+				"timer_id": stringProp("Timer id."),
+			}, []string{"timer_id"}),
+		},
+		{
+			Name:        "timer_resume",
+			Description: "Resume one paused timer owned by the caller.",
+			InputSchema: objectSchema(map[string]any{
+				"timer_id": stringProp("Timer id."),
+			}, []string{"timer_id"}),
+		},
+		{
+			Name:        "timer_list",
+			Description: "List pending and paused timers owned by the caller.",
+			InputSchema: objectSchema(nil, nil),
+		},
 		{
 			Name:        "scratchpad_list",
 			Description: "List shared per-project scratchpad entries.",
@@ -311,7 +395,8 @@ func (s *Server) handleProtocolMethod(callerID, method string, params json.RawMe
 			"capabilities": map[string]any{
 				"tools": map[string]any{"listChanged": false},
 			},
-			"serverInfo": serverInfo,
+			"serverInfo":   serverInfo,
+			"instructions": serverInstructions,
 		}
 		return result, true, 0, "", nil
 

internal/mcp/protocol_test.go

@@ -36,6 +36,13 @@ func TestInitializeReturnsCapabilities(t *testing.T) {
 	if caps["tools"] == nil {
 		t.Fatalf("tools capability missing: %+v", caps)
 	}
+	// patterm-specific orientation: clients show this to the underlying
+	// LLM, so it's our primary hook for steering vendor TUIs (codex in
+	// particular) toward the MCP tool surface instead of shell-ing out.
+	instructions, ok := parsed.Result["instructions"].(string)
+	if !ok || instructions == "" {
+		t.Fatalf("instructions missing or wrong type: %+v", parsed.Result)
+	}
 }
 
 func TestInitializedNotificationSuppressesResponse(t *testing.T) {

internal/mcp/tools.go

@@ -88,6 +88,13 @@ type ToolHost interface {
 	SendMessage(callerID, targetID, message string) error
 	RequestHumanAttention(callerID, processID, reason string) error
 	TimerWait(callerID string, seconds float64, label string) (string, error)
+	TimerSet(callerID string, args TimerSetArgs) (TimerHandle, error)
+	TimerFireWhenIdleAny(callerID string, args TimerFireWhenIdleArgs) (TimerFireWhenIdleResponse, error)
+	TimerFireWhenIdleAll(callerID string, args TimerFireWhenIdleArgs) (TimerFireWhenIdleResponse, error)
+	TimerCancel(callerID, id string) error
+	TimerPause(callerID, id string) error
+	TimerResume(callerID, id string) error
+	TimerList(callerID string) ([]TimerInfo, error)
 
 	// Scratchpads.
 	ScratchpadList() ([]scratchpad.Entry, error)
@@ -111,6 +118,13 @@ type ProcessInfo struct {
 	ExitCode        *int   `json:"exit_code,omitempty"`
 	IdleMS          int64  `json:"idle_ms,omitempty"`
 	Trusted         *bool  `json:"trusted,omitempty"`
+
+	// IdleState is the idle-detection classifier's current opinion:
+	// one of "idle", "working", "thinking", "permission", "error".
+	// Empty when the classifier has not yet evaluated this child
+	// (typically right after spawn) or when idle detection is disabled.
+	IdleState  string `json:"idle_state,omitempty"`
+	IdleReason string `json:"idle_reason,omitempty"`
 }
 
 // ProcessStatus is what get_process_status returns. Richer than
@@ -181,6 +195,63 @@ type SearchMatch struct {
 	Text   string `json:"text"`
 }
 
+// TimerSetArgs is the input for timer_set: a one-shot delay timer that
+// delivers Body to the owning agent as a fresh user turn when it fires.
+// OwnerProcessID is optional — when empty the caller's own process_id
+// is used (matching Solo's "bound agent" semantics). Top-level
+// orchestrators (no caller identity) must set OwnerProcessID
+// explicitly.
+type TimerSetArgs struct {
+	Body           string  `json:"body"`
+	Label          string  `json:"label,omitempty"`
+	Seconds        float64 `json:"seconds"`
+	OwnerProcessID string  `json:"owner_process_id,omitempty"`
+}
+
+// TimerFireWhenIdleArgs is the input for timer_fire_when_idle_any /
+// timer_fire_when_idle_all. Watched lists process_ids to monitor.
+// MaxWaitSeconds bounds how long the timer can stay pending before
+// firing anyway (0 = no max wait, fire only when the idle condition is
+// met). OwnerProcessID: see TimerSetArgs.
+type TimerFireWhenIdleArgs struct {
+	Body           string   `json:"body"`
+	Label          string   `json:"label,omitempty"`
+	Watched        []string `json:"watched"`
+	MaxWaitSeconds float64  `json:"max_wait_seconds,omitempty"`
+	OwnerProcessID string   `json:"owner_process_id,omitempty"`
+}
+
+// TimerHandle is the response for timer_set.
+type TimerHandle struct {
+	ID string `json:"timer_id"`
+}
+
+// TimerFireWhenIdleResponse covers timer_fire_when_idle_any /
+// timer_fire_when_idle_all. When every watched process is already idle
+// at registration time, idle_all returns Status="already_satisfied"
+// and ID="" — no timer is created (matches Solo). idle_any returns
+// AlreadyIdle so the caller can see which processes were excluded from
+// satisfaction.
+type TimerFireWhenIdleResponse struct {
+	ID          string   `json:"timer_id,omitempty"`
+	Status      string   `json:"status"` // "pending" | "already_satisfied"
+	AlreadyIdle []string `json:"already_idle,omitempty"`
+	WaitingOn   []string `json:"waiting_on,omitempty"`
+}
+
+// TimerInfo is one row in the timer_list response.
+type TimerInfo struct {
+	ID                string   `json:"timer_id"`
+	Label             string   `json:"label,omitempty"`
+	Body              string   `json:"body,omitempty"`
+	Kind              string   `json:"kind"`             // "delay" | "idle_any" | "idle_all"
+	Status            string   `json:"status"`           // "pending" | "paused"
+	OwnerID           string   `json:"owner_process_id"`
+	WatchedIDs        []string `json:"watched,omitempty"`
+	FiresAtUnixMS     int64    `json:"fires_at_unix_ms,omitempty"`
+	PausedRemainingMS int64    `json:"paused_remaining_ms,omitempty"`
+}
+
 // PortSighting matches the per-child store in internal/app.
 type PortSighting struct {
 	Port        int    `json:"port"`
@@ -575,6 +646,82 @@ func callTool(h ToolHost, callerID, method string, params json.RawMessage) (any,
 		}
 		return map[string]string{"timer_id": id}, 0, "", nil
 
+	case "timer_set":
+		var p TimerSetArgs
+		if err := unmarshalParams(params, &p); err != nil {
+			return nil, codeInvalidParams, err.Error(), nil
+		}
+		h2, err := h.TimerSet(callerID, p)
+		if err != nil {
+			return mapToolError(err)
+		}
+		return h2, 0, "", nil
+
+	case "timer_fire_when_idle_any":
+		var p TimerFireWhenIdleArgs
+		if err := unmarshalParams(params, &p); err != nil {
+			return nil, codeInvalidParams, err.Error(), nil
+		}
+		resp, err := h.TimerFireWhenIdleAny(callerID, p)
+		if err != nil {
+			return mapToolError(err)
+		}
+		return resp, 0, "", nil
+
+	case "timer_fire_when_idle_all":
+		var p TimerFireWhenIdleArgs
+		if err := unmarshalParams(params, &p); err != nil {
+			return nil, codeInvalidParams, err.Error(), nil
+		}
+		resp, err := h.TimerFireWhenIdleAll(callerID, p)
+		if err != nil {
+			return mapToolError(err)
+		}
+		return resp, 0, "", nil
+
+	case "timer_cancel":
+		var p struct {
+			TimerID string `json:"timer_id"`
+		}
+		if err := unmarshalParams(params, &p); err != nil {
+			return nil, codeInvalidParams, err.Error(), nil
+		}
+		if err := h.TimerCancel(callerID, p.TimerID); err != nil {
+			return mapToolError(err)
+		}
+		return "ok", 0, "", nil
+
+	case "timer_pause":
+		var p struct {
+			TimerID string `json:"timer_id"`
+		}
+		if err := unmarshalParams(params, &p); err != nil {
+			return nil, codeInvalidParams, err.Error(), nil
+		}
+		if err := h.TimerPause(callerID, p.TimerID); err != nil {
+			return mapToolError(err)
+		}
+		return "ok", 0, "", nil
+
+	case "timer_resume":
+		var p struct {
+			TimerID string `json:"timer_id"`
+		}
+		if err := unmarshalParams(params, &p); err != nil {
+			return nil, codeInvalidParams, err.Error(), nil
+		}
+		if err := h.TimerResume(callerID, p.TimerID); err != nil {
+			return mapToolError(err)
+		}
+		return "ok", 0, "", nil
+
+	case "timer_list":
+		ts, err := h.TimerList(callerID)
+		if err != nil {
+			return mapToolError(err)
+		}
+		return ts, 0, "", nil
+
 	case "scratchpad_list":
 		entries, err := h.ScratchpadList()
 		if err != nil {

internal/preset/preset.go

@@ -40,9 +40,42 @@ type Preset struct {
 	Shell bool `json:"shell,omitempty"`
 
 	// Agent-only. SPEC §10.
-	MCPInjection     *MCPInjection `json:"mcp_injection,omitempty"`
-	ReadySignal      *ReadySignal  `json:"ready_signal,omitempty"`
-	ChromeTrimHints  []string      `json:"chrome_trim_hints,omitempty"`
+	MCPInjection     *MCPInjection  `json:"mcp_injection,omitempty"`
+	ReadySignal      *ReadySignal   `json:"ready_signal,omitempty"`
+	ChromeTrimHints  []string       `json:"chrome_trim_hints,omitempty"`
+	IdleDetection    *IdleDetection `json:"idle_detection,omitempty"`
+}
+
+// IdleDetection configures steady-state idle classification for an
+// agent preset. Independent of ReadySignal (which is startup-only).
+// All fields are optional; when the whole block is nil the runtime
+// falls back to output_activity with a 2s threshold.
+//
+// Strategy selects the primary signal:
+//   - "output_activity": ms since last PTY output (Claude, OpenCode).
+//   - "osc_title_stability": ms since last OSC 0/2 title change
+//     (Codex, Amp — title changes mean activity).
+//   - "osc_title_status": substring-match the current title against
+//     TitleStatusMap (Gemini — title carries a status word).
+//
+// Promoter patterns are applied on top of the strategy. They run
+// against the recent ring-buffer tail; the first match wins in
+// error > permission > thinking precedence and promotes the state
+// over whatever the strategy returned.
+type IdleDetection struct {
+	Strategy        string `json:"strategy,omitempty"`
+	IdleThresholdMS int    `json:"idle_threshold_ms,omitempty"`
+
+	// TitleStatusMap maps a (case-insensitive) substring of the OSC
+	// title to a state. Only meaningful for "osc_title_status".
+	// Allowed values: "idle", "working", "thinking", "permission", "error".
+	TitleStatusMap map[string]string `json:"title_status_map,omitempty"`
+
+	// Output regex promoters. Compiled at load time; bad patterns are
+	// surfaced as warnings and skipped.
+	PermissionPatterns []string `json:"permission_patterns,omitempty"`
+	ThinkingPatterns   []string `json:"thinking_patterns,omitempty"`
+	ErrorPatterns      []string `json:"error_patterns,omitempty"`
 }
 
 // MCPInjection covers the strategies SPEC §10 enumerates plus
@@ -196,6 +229,15 @@ func ensureDefaults(base string) error {
   "argv": ["claude"],
   "mcp_injection": { "kind": "flag", "flag": "--mcp-config" },
   "ready_signal": { "idle_ms": 1000 },
+  "idle_detection": {
+    "strategy": "output_activity",
+    "idle_threshold_ms": 2000,
+    "permission_patterns": [
+      "Do you want to proceed\\?",
+      "❯ 1\\. Yes",
+      "1\\. Yes, and don't ask"
+    ]
+  },
   "chrome_trim_hints": [
     "^Welcome to Claude Code",
     "^/help for help",
@@ -220,6 +262,10 @@ func ensureDefaults(base string) error {
     "format": "toml"
   },
   "ready_signal": { "idle_ms": 1000 },
+  "idle_detection": {
+    "strategy": "osc_title_stability",
+    "idle_threshold_ms": 2000
+  },
   "chrome_trim_hints": [
     "^OpenAI Codex",
     "^\\s*model:",
@@ -243,6 +289,10 @@ func ensureDefaults(base string) error {
     "var": "OPENCODE_CONFIG_CONTENT"
   },
   "ready_signal": { "idle_ms": 1000 },
+  "idle_detection": {
+    "strategy": "output_activity",
+    "idle_threshold_ms": 2000
+  },
   "chrome_trim_hints": [
     "^\\s*█",
     "^\\s*opencode v",
@@ -250,14 +300,6 @@ func ensureDefaults(base string) error {
     "^\\s*>_"
   ]
 }
-`,
-		},
-		{
-			"presets/processes/shell.json",
-			`{
-  "name": "shell",
-  "argv": ["__SHELL__"]
-}
 `,
 		},
 	}
@@ -269,15 +311,7 @@ func ensureDefaults(base string) error {
 		if err := os.MkdirAll(filepath.Dir(full), 0o700); err != nil {
 			return err
 		}
-		body := d.body
-		if strings.Contains(body, "__SHELL__") {
-			shell := os.Getenv("SHELL")
-			if shell == "" {
-				shell = "/bin/sh"
-			}
-			body = strings.ReplaceAll(body, "__SHELL__", shell)
-		}
-		if err := os.WriteFile(full, []byte(body), 0o600); err != nil {
+		if err := os.WriteFile(full, []byte(d.body), 0o600); err != nil {
 			return err
 		}
 	}

internal/vt/emulator.go

@@ -57,6 +57,11 @@ type Emulator interface {
 	// ActiveScreen reports whether we are on the primary or alternate buffer.
 	ActiveScreen() (Screen, error)
 
+	// Title returns the most recently set window title (OSC 0/2). Returns
+	// an empty string if no title has been set. Used by idle detection
+	// for the osc_title_stability and osc_title_status strategies.
+	Title() (string, error)
+
 	// ScrollViewportTop moves the viewport to the top of the scrollback.
 	ScrollViewportTop() error
 

internal/vt/ghostty.go

@@ -544,6 +544,27 @@ func (e *GhosttyEmulator) Cursor() (CursorState, error) {
 	return CursorState{Col: uint16(col), Row: uint16(row), Visible: bool(visible)}, nil
 }
 
+// Title returns the most recent window title set by OSC 0/2 escape
+// sequences. The libghostty-vt API hands back a borrowed pointer that
+// stays valid only until the next vt_write/reset, so we copy out to a
+// Go string under the same mutex that gates writes. An empty string
+// (len=0) means no title has been set.
+func (e *GhosttyEmulator) Title() (string, error) {
+	e.mu.Lock()
+	defer e.mu.Unlock()
+	if e.closed {
+		return "", errors.New("vt: emulator closed")
+	}
+	var s C.GhosttyString
+	if rc := C.ghostty_terminal_get(e.term, C.GHOSTTY_TERMINAL_DATA_TITLE, unsafe.Pointer(&s)); rc != C.GHOSTTY_SUCCESS {
+		return "", fmt.Errorf("vt: get title failed: %s", ghosttyResultStr(rc))
+	}
+	if s.ptr == nil || s.len == 0 {
+		return "", nil
+	}
+	return C.GoStringN((*C.char)(unsafe.Pointer(s.ptr)), C.int(s.len)), nil
+}
+
 func (e *GhosttyEmulator) ActiveScreen() (Screen, error) {
 	e.mu.Lock()
 	defer e.mu.Unlock()

internal/vt/ghostty_nocgo.go

@@ -24,6 +24,7 @@ func (e *GhosttyEmulator) SerializeVT() ([]byte, error)    { return nil, errStub
 func (e *GhosttyEmulator) StyledScreenVT() ([]byte, error) { return nil, errStub }
 func (e *GhosttyEmulator) Cursor() (CursorState, error)    { return CursorState{}, errStub }
 func (e *GhosttyEmulator) ActiveScreen() (Screen, error)   { return 0, errStub }
+func (e *GhosttyEmulator) Title() (string, error)          { return "", errStub }
 func (e *GhosttyEmulator) ScrollViewportTop() error        { return errStub }
 func (e *GhosttyEmulator) ScrollViewportBottom() error     { return errStub }
 func (e *GhosttyEmulator) ScrollViewportDelta(int) error   { return errStub }