loopd: subagent spawn on the background slot

Co-Authored-By: ornith-1.5-35b-a3b <agent@wntrmute.dev>

docs/implementer-log.md

@@ -6,6 +6,7 @@ reviewer adds findings under "Reviews" once per milestone.
 
 | Task | Date | Status | Gate runs | First gate | Deviations | Notes | Model |
 |---|---|---|---|---|---|---|---|
+| M6a/03-loopd-subagent-spawn | 2026-10-03 | done | 1 | pass | none | Copied the given test `crates/loopd/tests/subagent_spawn.rs` from `docs/plans/M6a/files/`. Step 2 (`cargo build --tests --keep-going -p loopd`) failed as expected: E0560 `Runtime` has no field `ctx`. Added `pub ctx: Option<&'a crate::channel::Context>` to `Runtime` in turn.rs R1, filled it `Some(&*ctx)` in channel.rs C1 (live turn), `Some(ctx)` in the cron and heartbeat literals in scheduler.rs S1/S2 (both already held an immutable borrow of `ctx`), and `None` in the seven test literals: support/turn.rs, compact_turn.rs (the helper and the inline one), session_slot.rs, end_to_end.rs, turn_broker.rs, tool_summary_call.rs. Replaced the stub `run` in subagent.rs with the full spawn: build `sub-<parent>-<n>` from the already-consumed `call_id.0`, overflow returns `denial_message(InvalidArguments, Some(...))` naming subagent-id without holding, then `rt.ctx` `None` and `hold Busy` and block/baseline/create errors each return `failed`, a `Create` `Exists` falls back to `open`, the child runs one `run_turn` plus `finish_turn` on `slots.background`, memory is updated, `untrusted` is read before the hold drops, and the result is `{id} completed\n{summary}` or `{id} failed`, all `Public`. Added `RUNBOOK` and the use lines the body needs. Appended `## subagent-id` to docs/runbook.md via the task's `{ echo; cat ...; }` (grep -c prints 1). `cargo check -p loopd` clean, no warning; `wc -l` prints 185 subagent.rs, 335 turn.rs, 301 channel.rs, 304 scheduler.rs, all under 500. All suites pass: `subagent_spawn` 5, `subagent_dispatch` 3, `subagent_registry` 2, `session_slot` 4, `tool_summary_call` 6, `compact_turn`, `channel`. `make gate` printed `gate: ok` with 1585 tests (1580 before, plus 5). No new dependency, so `Cargo.lock` unchanged. | ornith-1.5-35b-a3b |
 | M6a/02-loopd-subagent-dispatch | 2026-10-03 | done | 1 | pass | none | Copied the given test `crates/loopd/tests/subagent_dispatch.rs` from `docs/plans/M6a/files/`. Step 2 (`cargo build --tests --keep-going -p loopd`) failed as expected: E0599 no `Subagent` variant and E0425 no `child_id`. Added `Dispatch::Subagent { task }` in tools.rs D1, routed `subagent` in `dispatch` and `dispatch_call_tool` (D2/D3) via `crate::subagent::dispatch_args`, added the `Subagent` arm in `run_call` (turn.rs T1) calling the stub `run`, and appended `child_id`, `parse_task`, `dispatch_args` and the stub `run` to subagent.rs. `parse_task` walks every key (unknown field), requires `task`, demands a string task and a JSON object arguments, and maps errors to `denial_message(InvalidArguments, Some(&detail))`. `run` stays a compiling stub, no spawn, no `Runtime.ctx`. `cargo check -p loopd` clean, no warning; `wc -l` prints 76 for subagent.rs, 477 for tools.rs, 332 for turn.rs, all under 500. All suites pass: `subagent_dispatch` 3, `subagent_registry` 2, `tools` 11. `make gate` printed `gate: ok` with 1580 tests (1577 before, plus 3). No new dependency, so `Cargo.lock` unchanged. | ornith-1.5-35b-a3b |
 | M6a/01-loopd-subagent-registry | 2026-10-03 | done | 1 | pass | none | Copied the given test `crates/loopd/tests/subagent_registry.rs` from `docs/plans/M6a/files/`. Created `crates/loopd/src/subagent.rs` with `SUBAGENT` and `schema()` (a non-core `ToolSchema`, no `parse_task`/`dispatch_args`/`run`). Added `pub mod subagent;` after `pub mod sse;` in lib.rs (rustfmt order). Edited `m5c()` in tools.rs to push the non-core `subagent` entry after the `ask_memory` insert at index 3, so `ask_memory` stays at 3 and `subagent` is last among the non-core names; `core_schemas()` still returns only the six core names and `find("subagent")`/`find("isolated")` both hit it. `cargo check -p loopd` clean, no warning; `wc -l` prints 22 for subagent.rs and 470 for tools.rs, both under 500. All suites pass: `subagent_registry` 2, `ask_memory_tool`, `tools` 11, `baseline` 6. `make gate` printed `gate: ok` with 1577 tests (1575 before, plus 2). No new dependency, so `Cargo.lock` unchanged. | ornith-1.5-35b-a3b |
 | M5d2/11-loopd-heartbeat-failed-last | 2026-10-03 | done | 1 | pass | none | Copied the given test `crates/loopd/tests/scheduler_heartbeat_failed.rs` from `docs/plans/M5d2/files/`. Applied H1 in `crates/loopd/src/scheduler.rs` inside `heartbeat`: after the quoted block (found exactly once) the bookkeeping is now unconditional — `let _ = compact::finish_turn(...)` instead of binding `outcome`, then `drop(held)`, then `self.last_heartbeat = Some(now)` and `work.push(TickWork::Heartbeat)` with no `if outcome.is_ok()`. The unreadable-prompt and busy `return` paths stay untouched and never reach H1; cron `Finished`/`Failed` and the job arm are unchanged. `cargo test -p loopd --test scheduler_heartbeat_failed` failed first (`left: []`, `right: [Heartbeat]`), then passed with `scheduler_heartbeat` (6). `cargo check -p loopd` clean, no warning. `make gate` printed `gate: ok` with 1575 tests (1574 before, plus 1). No new dependency, so `Cargo.lock` unchanged. | ornith-1.5-35b-a3b |