- Strict steer additionally requires an OPEN STEP: between steps the
loop may be awaiting its continuation/turn-stop checkpoints, where
pending steering was already folded and a terminal stop discards a
later arrival. A message accepted during an open step is drained and
recorded at that step's settlement before any terminal decision, so
the acknowledged-then-discarded window is closed. New keyless test
holds agent/turn-stop open and pins the rejection.
- tool-subagent-control README: distinguish synchronous not-delivered
errors from started-Task failures (unknown/foreign/descriptor-less
ids settle the started Task as failed), and drop the claim that the
completion notice carries the child's response.
- Wire the control service and send_message tool into every shipped
composition with a resumable provider and background enabled
(headless-agent, tui-agent, and the SDK helper's subagent feature
base resources); jsonrpc-agent disables background and is unchanged.
- Resolve the send_message availability check in the CALLER's tool
scope so a restriction that removes the follow-up tool from one
agent also blocks that agent's continuable start.
- Control-service disposal now cancels live activations and awaits
producer settlement instead of stranding them: TaskService keeps
producer Tasks across a reload, so the disposing service aborts each
activation-owned controller, resolves its terminal gate (the
effect-scoped onTaskDone listener is already gone), and awaits done.
A new test kills a mid-start activation through HMR disposal.
- Strict steer now rejects the two windows where an acknowledged
message would be silently dropped: the closed-turn durability-flush
window (status still running, loop strands drained steering) and a
committed structured capture (terminal turn-stop discards late
steering). Seam JSDoc, catalog doc, README, and the Agent Note
bilingual pair state the tightened contract; new keyless tests pin
both rejections.
- Continuable background delegation now fails loud when the advertised
send_message tool is not registered, instead of starting a durable
child the model cannot continue. The acp-agent example already loads
the control tool; the tool-catalog boot recipe is unaffected because
capability wording is harvested at mount.
macOS realpaths temp dirs into /private. The normalizer only stripped
the /private prefix for TMPDIR under /var; a TMPDIR under /tmp (any
explicitly relocated temp root) left one side canonicalized and the
comparison failing. Accept both shapes.
Implement the continuable background subagents RFC: a durable child
session with a series of Task-backed activations, each disposing its
run before the Task settles.
- dsh-subagent: rename SubagentRun.sendMessage to strict steer, drop
run-level resume, add SubagentProvider.resume dispatch via
SubagentService.resume, the continuation start field, and the
versioned model-hidden subagent/descriptor session event.
- dsh-subagent-inprocess/-spawn/-fork: publish the control-allocated
child id, append the descriptor inside the initial turn, implement
cold resume from the child's own transcript under the live parent
scope, and strict running-only steer.
- dsh-subagent-control (new): SubagentControlService owning stable
child ids, descriptor snapshot/fold/authorization, Task-backed
activation with settle-then-dispose ordering, the process-local
active-run association, and steer-or-resume sendMessage routing.
- dsh-tool-subagent: background route branches on the provider's
resume capability (continuable via the control service; one-shot
task for ACP), returning both child and task ids.
- dsh-tool-subagent-control (new): the globally named send_message
tool rendering steered/started routes.
Keyless coverage spans Task ownership and disposal ordering, running
delivery, cold follow-up, descriptor rejection and rollback, known-id
reconstruction, kill during lookup, admission races, and a new
subagent-continuable ACP snapshot scenario.
The shipped-roster change made @deepseek-ai/dsh-tool-session-query a
default row of the shared base.cordis.yml, so the TUI and Web surfaces
put the five session-search tools in front of the model. That
contradicts the recorded opt-in stance for the model-facing session
query consumer; the ACP example remains the mounted reference.
Remove the row from the shared base, the now-dangling disabled patch in
the opt-in core-web profile, and the workspace dependency. The
ctx.sessionQuery index stays: the TUI's /resume and the Web content
search consume it directly. Both shipped-composition tests now pin the
20-tool catalog.
goalChangeRef returned the full GoalSnapshot for every snapshot operation,
so foldGoal(...).lastRef and the goal/changed notification ref carried
objective, phase, and maxGoalRounds fields instead of the declared
GoalRef { id, revision }. Only the clear tombstone was bare. Emit an exact
{ id, revision } ref for snapshot changes and pin the contract with a
regression test covering create/edit/block notifications and the fold.
CI caught a startup race the PTY smoke stages naturally: a TUI /exit typed
while watchPersonalPatches is still opening its watcher disposes the tree,
and the HMR effect registration rejects with INACTIVE_EFFECT, crashing an
app that exited exactly as asked. Return a no-op disposer for that exact
code; every other registration failure still propagates (covered both ways).
ds-review-bot round 1 on the DSH-home integration:
- generated wrappers now inject the services their manifest needs (skills/
tools beside loader), and loadPreparedRepository rejects a wrapper fiber
that settles anything but ACTIVE — a composition missing a required
service fails the repository transaction instead of committing an ACTIVE
row over a silently PENDING child (critical finding)
- the github: source ref segment excludes '#', so 'a#b' refs fail at the
config parser with the promised syntax instead of inside pnpm
- watchPersonalPatches re-reads the include's non-patch options per refresh
instead of a registration-time snapshot
- the TUI smoke's cache-seeded wrapper is produced by the real
prepareDshPlugin (cache LAYOUT stays a deliberate external pin)
- new Loader integration test drives a live repositories update through
entry.update: generation swap, old skills removed, failed candidate
rolled back to the previous generation
ds-review-bot round 1 on the repository-plugin runtime:
- a manifest-declared skill root absent or non-directory in the installed
package now fails the plugin load (skill-local treats a missing root as
legitimately empty, which silently mounted a skill-less plugin)
- includeDefaultRoots: false no longer inherits $DSH_BUNDLED_SKILL_DIR, so
isolated repository providers see only their explicit roots
- prepared wrapper baseUrl schema requires the file: scheme, failing hostile
URLs at the declared validation boundary
- preparedPath reuses format.ts's isOutside; SERVER_NAME_PATTERN is exported
and pinned equal to dsh-mcp-client's, with the restatement justified (the
prepare bin keeps a zod-only module graph); the unexplained `as never`
cast now carries its schemastery rationale
- the import-free wrapper assertion also rejects dynamic import(
- the headless fixture wrapper is regenerated by the real prepareDshPlugin
and a drift test pins fixture == generator output
- prepareDshPlugin JSDoc states the non-atomic publish repair contract
ds-review-bot round 1: boot() now throws `host preparation failed` when
prepare() rejects before any config-tree entry mounts (the plugin-tree label
overstated), and the new hygiene gate verify-vendored-links pins the
linkWorkspacePackages fix — every vendored package name in pnpm-lock.yaml
must resolve to a workspace link with no registry copy alongside.
The timeout promise's executor runs synchronously while the race is
constructed, so the timer is always assigned; the undefined check was a
dead branch the per-file coverage gate rejected.
The fork button on a stopped assistant message was inert. Frozen
interrupted nodes carry a flow-ordering seq of turnEnd.seq - 0.9, and
session.fork takes a non-negative integer on the wire, so every such
request was rejected as invalid-params before reaching the host — where
an aborted turn's logged turn/end has always made it forkable.
SessionsService.fork floors atSeq at the wire boundary. Flooring stays
inside the anchor's own turn (every turn opens with turn/start), so the
host's first-turn/end-at-or-after cut still closes on that turn.
The PTY capture does continue past the terminal-takeover bytes with the
fatal diagnostic; only the reset never follows. State that precisely in
both notes.
Document on FailLoudProcess.exit that callers treat it as the end of the
run, matching how the release path already relies on it.
Review of the previous commit found two defects in the release path, both
reproduced against the implementation:
- The timeout guarding a never-settling release was unref'ed. An
unhandledRejection listener suppresses Node's default fatal exit, so with
nothing else referenced the process reached an empty event loop and exited
0 on the very failure it was reporting. Keep the timer referenced and clear
it once the race settles.
- The handler uninstalled itself before awaiting the release. A second
concurrent rejection then became uncaught and killed the process
mid-teardown, stranding exactly the terminal state this restores. Replace
the uninstall with a latch: the first rejection is the reported one, and
later rejections (teardown's own included) fall through to the pending exit.
Add the PTY regression the fake-process tests cannot express: boot the shipped
tree over a fixture whose llm-pi-ai providers value is list-shaped, expect exit
1, and assert the captured bytes carry both the diagnostic and ESC[?2004l.
Against the pre-fix source the stream ends at ESC[?2004h ESC[>7u ESC[?u ESC[c
with no reset and the case fails, so it pins the actual bug.
Split the two-shape formatting test into one install per case; a latched
handler reports once by design.
Dudu-0223