The executor collapse landed without telling the model it exists. Every
tool contributes its own guidance section naming its tool, none of them
qualify how that tool is reached, and they all render before the SDK
(orders 100-199 against SDK_SECTION_ORDER 150), so the prompt said "Use
the read tool" eleven times and never said only run_code is callable.
A real session shows the consequence: the model emitted a native call,
read `unknown tool "read"` for a tool the same prompt declares, and
concluded the deployment was inconsistent rather than routing through
run_code.
The registry now contributes `tools:code-only` at order 99 -- ahead of
the guidance band -- stating the rule, registered wherever `tools:sdk`
is and rendering empty outside an effective `code`. `both` renders it
empty because its native calls do execute, which is also why
both-mode-turn no longer shares code-mode-turn's expected prompt. The
denial itself now names the route back, since a bare UNKNOWN_TOOL for a
declared tool is what misled the model.
LegGasai noted that filtering prompt sections by tool:* prefix is a poor
heuristic: it conflates section naming convention with presentation
semantics and would incorrectly drop tool:structured_output. The executor
collapse already enforces the boundary — a model-direct native call is
rejected as UNKNOWN_TOOL regardless of what the prompt says, so filtering
the prompt adds no security and only risks losing useful guidance.
The tool:read/tool:bash/etc sections describe capability usage patterns
that apply to both native and code presentations, and keeping them does
not reopen the native direct-call path because the executor blocks it.
Only `docs/module-graph.*` conflicted, from master's new
`@deepseek-ai/dsh-message-feedback` package; generated, so taken from master
and regenerated.
Hook bypassed as before: the staged-pairing check hands an archived note path
to `verify-translation-pairing`. The full-corpus gate passes.
Generated docs (`module-graph`, `event-producer-consumer`) taken from master
and regenerated. The `cordis-inspect-jsdoc` golden likewise: master's copy is
the base, and this branch's `presentAs` per-scope rewording is re-applied on
top, since `cordis_inspect` renders that JSDoc into model-visible output.
Hook bypassed as before: the staged-pairing check hands an archived note path
to `verify-translation-pairing`. The full-corpus gate passes.
One assistant message carrying several subagent calls now overlaps them
under the rolling pool (maxParallelToolCalls) instead of serializing each
foreground delegation behind an exclusive barrier; results still commit
in model order.
tool-subagent declares isConcurrencySafe: () => true for every call form:
children work in their own sessions, a run never mutates the parent
session, and sibling workspace coordination already belongs to the model
for background, continuable, and workflow children. The former
serial-execution pin flips to a parallel pin plus a gated genuine-overlap
test, and the authored subagent-parallel snapshot pins the assembled
transcript (tool/call, tool/call, tool/result, tool/result) with
interchangeable twin children, which replay race-free under the existing
first-call binding and createdAt harvest ordering.
The provider seam now requires concurrent starts and continuable
preparations for distinct children to isolate operation-local state,
cancellation, settlement, and cleanup; a continuable gate test proves a
cancelled preparation leaves no Agent or durable Session while its
sibling persists independently.
Closesdeepseek-harness/deepseek-harness#1489
list_agents called a storage-only continuable child 'complete', inviting
parents to poll it for a result that will never appear there. Rename the
status to 'ready' — resumable, not terminal — and rewrite the tool
description to say a parent is told when a child finishes, so listing is
for recall, not polling.
A continuable child that stopped without reporting — an error, a token
ceiling, cancellation, teardown — left its parent nothing to act on.
The continuation manager now delivers an unconditional settlement
notice to the durable direct parent before releasing ownership, folding
consumed work (foldConsumedWork supersedes findLastMessageTurnEnd) so a
claimed-but-unrun prompt reads as aborted rather than completed, waking
an idle parent, steering a busy one, and never waking a closing tree.
A continuable child could finish without giving its parent a usable
result: the report tool was described as optional and its default
delivery stayed quiet. Install the child-scoped report obligation (tool
description plus the tool:report prompt section), default reportDelivery
to wakeup, bind shipped subagent_fork delegations to one-shot so fork
keeps its prefix reuse, and extend the ACP snapshot harness to pin child
system prompts alongside child tool schemas.
The pre-commit staged-pairing hook was bypassed for this merge: master's
`docs(notes): archive superseded dsh run decision` (7ee9e16001) arrives as a
rename into `.agents/notes/archived/`, and the hook hands that path to
`verify-translation-pairing`, which correctly refuses an archived note as
out-of-corpus. The full-corpus gate passes (851 pairs).
The vendored Cordis rescope (`cordis` -> `@deepseek-ai/cordis`,
`@cordisjs/plugin-*` -> `@deepseek-ai/cordis-plugin-*`) collided with this
branch's two added edges. Resolved by taking master's package names and
keeping this branch's additions: `@deepseek-ai/dsh-agent` and
`@deepseek-ai/dsh-system-prompt` peers, the `scopeChainOf` import, and the
`agent/created` type-only edge.
`docs/event-producer-consumer.*` is generated; regenerated after the merge
rather than hand-resolved.
Moving the agent plane behind presets left two readers on the wrong side of
the host/agent line.
`dsh-token-meter` was disabled on the host and mounted inside each preset's
`compaction` realm, but its three projection units register into the
process-wide `sessionProjections` table. A unit registered from one preset
answers for every session, so whether a `minimal` session showed a context
meter depended on whether some other session had mounted `standard` since
boot, and a process that only ever ran `minimal` showed none. The meter takes
no configuration, keys every fold by Session, and registers no tool or prompt
section, so it returns to the host composition and leaves the presets'
`isolate` map; the realm and `compact-basic` stay, because what a preset
chooses is whether its agent compacts, not whether its tokens are counted.
Nothing named an agent that joined no preset. The join is a scope-parent link,
and without it the tools, prompt-section, and skill views resolve the empty
global layer: the agent publishes, the turn runs, and the model receives
nothing. `AgentPresets` now logs one warning per such agent while a roster is
configured, and the invariant companion fails outright — at
`system-prompt/assemble` rather than at publication, because an unjoined agent
is legal until it addresses a model and `recompose` binds exactly such an
agent. The warning stays advisory: a synchronous `agent/created` throw vetoes
publication, and the ACP bridge, SDK server, and headless bundle all create an
unjoined agent today.
Three limits are recorded rather than fixed: projection key presence is not a
per-session capability signal, a superseded standing generation is never
reclaimed, and a `cordis_mount` temporary plugin belongs to the composition
rather than the session that mounted it.
Fixes#2203
