@bojackduy/opencode-loopd 1.10.1 → 1.10.4

package/CHANGELOG.md

@@ -1,5 +1,20 @@
 # Changelog
 
+## Unreleased — Command sessions replace opencode-pty + sentinel
+
+> `loopd_command_*` now covers everything `opencode-pty` (`pty_spawn/write/read/list/kill`) and `opencode-sentinel` (`sentinel_monitor/stop/list`) do — remove both plugins from config. Live-smoked on the real PTY backend (timeout, until-stop, pattern read, env, terminate+remove).
+
+### Added
+
+- **M1 pty parity** — `env` (values to host, names-only persistence), `timeout_seconds` (in-memory deadline → terminate path, `endReason=timeout`, always notifies), `shell:true` (`/bin/sh -c`), `pattern`+`ignore_case` log read (ANSI-stripped match, ReDoS guard), `terminate remove:true`, `endReason` (`exit|terminate|timeout|until|missing`) on every terminal transition.
+- **M2 sentinel watcher engine** — `watch_filter`/`watch_until`/`watch_ignore_case`/`watch_until_action (stop|keep)` at start, `loopd_command_watch` on running commands; 2s coalescing (20 lines/4KB), 100-lines/sec flood-suspend (process keeps running), 30-push budget; `until:stop` terminates with `endReason=until` and exactly one owner message; `loopd_command_await until:` wakes a goal once on match without stopping.
+- **M3 surface** — watch badges + endReason in Commands tab, watch counters in detail panel, SKILL.md watching guide with pty + sentinel migration tables, de-polled `loopd_command_get` docs.
+- **Owner auto-notify policy** — failure, lost-host `missing`, `timeout`, or ≥2min long-running success; `notify_on_exit` overrides.
+
+### Live proof
+
+- Scratch harness (real host + real service): timeout→`terminated/timeout`+notify; until→`terminated/until` with 1 notify and `until-matched` state; pattern read returns ANSI-original lines case-insensitively, rejects `(a+)+`; env value reaches process while state keeps only the name; terminate+remove in one call. 555 tests pass.
+
 ## 1.7.0 (2026-08-25) — Scheduled Intervals
 
 > Interval requeue for repetitive dialogues — one `/goal` covers `monitor deploy` `periodic tests` `keep going` without manual re-prompt. `5s` `createScheduleWorker` resurrects `complete → active` with `Scheduled tick N/M`.

package/README.md

@@ -144,12 +144,15 @@ The agent will clarify (what/where/how to verify) and then call `loopd_create_go
 
 Press **`<leader>o`** or open the command palette → **"Loop Dashboard"** (also `/loop`).
 
-Dashboard (NORMAL / INSERT `:`):
-
-- `j/k` — move selection
-- `g/G` — top / bottom
-- `o` — open child session (full transcript, native OpenCode view)
-- `p/r/R/x` — pause / resume / retry / clear selected goal
+Dashboard (NORMAL / INSERT `:`) — shared Goals/Commands tabs:
+
+- `Tab` — toggle Goals / Commands · `h` Goals · `l` Commands (`/commands` opens on Commands)
+- `j/k` — move selection · `g/G` — top / bottom (active tab only)
+- `o` — goals: open child session (native OpenCode view) · commands: close
+  popup, open fullscreen terminal page (route `opencode.loopd.terminal`)
+- `p/r/R/x` — pause / resume / retry / clear selected goal (Goals tab only)
+- Commands tab: `:new <command> [args...]` launch · `:interrupt` ·
+  `:terminate` · `:remove` · bare text + Enter writes stdin (owner-scoped)
 - `:` — insert mode → `:send <message>` to steer, `:force` / `:block` to finish manually
 - `?` — toggle help — help stays open while you type
 - `Ctrl+N` — back to NORMAL
@@ -157,6 +160,10 @@ Dashboard (NORMAL / INSERT `:`):
 
 > The dashboard traps all keys — even when open via the palette, the chat input cursor won't blink underneath.
 
+Fullscreen terminal page: raw typing goes straight to the PTY, `Ctrl+C`
+interrupts (never closes OpenCode), `Ctrl+]` detaches back to your session
+(the command keeps running — termination is explicit-only), paste is instant.
+
 ### 3. Inspect from the main agent (owner tools)
 
 The parent never needs the dashboard — it can ask:

package/commands/goal.md

@@ -6,6 +6,8 @@ Create a new background loop goal using the `loopd_create_goal` tool.
 
 A goal is a **contract**: `objective` (semantic requirements) + `checks` (deterministic host acceptance) + `agent`/`model`/`workspaceWrite`/`checkCwd`/`limits`. The host is the acceptance authority; the worker proposes completion and the host rejects it if `checks` fail (free retry <3, `blocked` after 3).
 
+If the user just wants to **run and watch/interact with a single process** (dev server, `test --watch`, REPL, log tail, one-off script) with no multi-turn AI reasoning involved, that's a `loopd_command_start` command session instead — lighter weight, no agent/checks/turns. Only proceed with this command when the request genuinely needs an autonomous AI worker.
+
 First, gather what you need to craft a good contract:
 - If the user gave a vague objective, ask 1–3 short clarifying questions (what to accomplish, where, and how they'll verify it — i.e., what `checks` should be).
 - If the user was specific, skip straight to creating it.