Translate ctx.bash.run() REJECTIONS into the SEARCH_* taxonomy. The seam
contract has run() reject for infrastructure failures (a pre-aborted
signal, an unusable/deleted session workdir, a missing shell); the bare
await let those escape as plain Errors, so the tool registry produced
isError results without the structured SearchError { name, code } the
package documents. A pre-aborted spec.signal now maps to SEARCH_ABORTED
and any other start failure to SEARCH_FAILED, original error chained as
cause. Covered by fake-executor tests for both branches plus real-executor
integration tests pinning the exact pre-aborted-signal and deleted-cwd
paths.
Two functional gaps in the search tools change:
- Enforce rawOutputMaxBytes on UNTRUNCATED inline stdout too. The cap was
only checked on the truncated->raw-spill path, so an executor retaining
more inline than the search cap (or a deployment lowering the cap below
the bash retention) could smuggle an over-cap parse through, contradicting
the documented SEARCH_RAW_OUTPUT_OVERFLOW contract. Covered by a new
over-cap-inline test.
- Load @deepseek-ai/dsh-timeout-policy in the coding-agent tree. The search
tools declare timeoutMs but nothing in the demo enforced it, so the
advertised 30s budget silently degraded to the bash executor's 60s
backstop. The keyless smoke boots the amended tree.
Implements docs/rfc/implemented/feature/2026-07-09-bash-backed-grep-glob-
discovery.md: model-facing glob/grep in a new @deepseek-ai/dsh-tool-fs-search
package, executing fixed ripgrep templates through ctx.bash.resolve/run —
not ctx.fs provider methods — so filesystem backends stay free of a search
contract and sandboxed/remote executors substitute cleanly. The tools never
call ctx.bash.start(); the tool layer owns quoting (one singleQuote safety
boundary), rg --json parsing, ItemRetainer/TextRetainer retention, and the
first tool-owned ctx.spillFiles.saveText() handoff (item-level retention the
generic post-execute spill policy cannot recover).
RFC amendments on the way to implemented/: a shared src/search-core.ts (the
SEARCH_* vocabulary + bash-run/raw-spill/spill plumbing was byte-identical
across both tools — the missed-extraction smell), and a snapshot-gap note:
wiring the acp-agent tree changes the assembled prompt, so goldens need a
keyed re-record; the spill notice text is pinned by unit tests instead and
only the coding-agent example ships the tools for now.
- spill-policy enforces the true cap invariant: it never emits a replacement
larger than maxInlineBytes. When the notice alone exceeds the cap (tiny cap or
long spill root) there is no within-cap replacement, so the inline result is
kept — the previous guard only compared against the original size and could
still return content over the cap for a large original. A within-cap
replacement is always smaller than the original, so this subsumes the earlier
check.
- Add the HMR-disposal test the conventions require for a new registration:
dispose the plugin fiber and assert oversized results stop being transformed
and nothing more is spilled (no leaked tools/post-execute listener on reload).
- spill-policy reserves the spill notice's byte cost inside maxInlineBytes, so
the replacement (preview + notice) never exceeds the documented model-facing
cap. When the notice alone fills the budget the preview is empty; when even a
notice-only replacement is not smaller than the original, the inline result is
kept (spilling would only add bytes).
- retention TextRetainer trims an oversized single suffix chunk to the last
suffixCap bytes on push, so tail/headTail retention stays bounded by suffixCap
instead of retaining and re-copying the whole chunk in finish() — this is the
spill preview path, which pushes the whole result as one chunk.
A reviewer read "TypeScript via host-side type-strip" and reasonably
asked what typing buys if nothing checks it — the group README never
said the annotations are advisory by design. The row now states it; the
rationale stays in the RFC and the enforcement story (per-dispatch
validateArgs) in the dsh-tools README.
- spill-policy validates maxInlineBytes as a non-negative integer at LOAD, so a
bad config fails the deployment instead of letting a negative value reach
TextRetainer and turn every oversized-result call into an isError.
- Document the spill seam vocabulary in docs/core-data-structures/spill.md
(SaveTextSpill/SpillOwner/SpillSource/SpillRef/SpillPath, verbatim + type-equiv
gated) and index it from core.md, matching the other capability seams.
Two [P1] review findings on the worker runtime:
- maxValueBytes gated and sliced the rendered fallback by UTF-16 code
units, so a multibyte string ("€€€€" under a 4-byte cap) crossed whole
and a truncated multibyte rendering could still run ~3x over budget.
New truncateUtf8Bytes cuts at code-point boundaries under a real byte
budget; prepareValue's fallback and the host's forged-error-text bound
both use it, and the VALUE_RENDER_SLACK comment drops its now-obsolete
"sliced by characters" wrinkle.
- The patched stream write dropped Node's optional encoding/callback
arguments, so a program awaiting flush completion
(write(chunk, resolve)) hung to the wall ceiling and misreported as a
timeout. The shim now fires the callback asynchronously once the chunk
is admitted — including for writes the exhausted budget drops.
Oversized plain-text tool results now spill to a session-scoped file and
return a bounded preview plus the spill path, so a verbose result stays
readable via `read` without consuming the next model request in full.
- dsh-spill: minimal SpillFiles seam (saveText → session-scoped SpillPath)
- dsh-spill-local: private 0700 session dirs, traversal-safe names, exclusive
owner-only writes
- dsh-spill-policy: tools/post-execute transformer; no-op unless maxInlineBytes
is set; skips read; best-effort on save failure (never turns a success into
an isError)
web_fetch is the showcase — no tool-specific spill code. The coding-agent
example loads the stack so its keyless Loader smoke guards the namespace-plugin
export shape. Snapshot gap for a transcript-visible web_fetch spill is recorded
in the RFC's Consequences (ACP replay is keyless and cannot hit the web).
The master merge added a code-runtime/ package row while this branch adds
the timeout/ row; together they push packages/README.md over its 605-word
ceiling. Condense the timeout/ row to the terse sibling style and raise the
ceiling 605->610 for the genuinely-new package group, mirroring how the
code-runtime work raised architecture.md's ceiling in the same spirit.
The zero-config enforcer no longer imports schemastery (its Config was
removed); knip flags the stale dependency. Remove it from the manifest
and sync the lockfile.
fetchTimeoutMs/searchTimeoutMs (default 30000) resolve to each tool's
ToolDefinition.timeoutMs, moving the budget's declaration home onto the
owning tool plugin and preserving per-tool deployment override without a
mistypable central tool-name map.
The enforcer now reads ctx.tools.get(exec.name).timeoutMs instead of a
free-text tool-name config map, so a mistyped name is impossible and the
tools/change warn-once apparatus is gone. exec.name always resolves in the
registry during dispatch, so there is no unknown-name path to warn about.
A tool declares its cooperative timeout budget on its own definition
rather than a deployment naming it in a central config map. The field
never reaches the model (schemas() whitelists name/description/parameters)
and defineTool rejects a non-positive-finite value at authorship.
Two findings from the GitHub review bot on the ready PR:
The tsdown two-entry build emitted the shared bootstrap module as a
lib/bootstrap-*.js chunk imported by both bundles, which the package.json
files whitelist (deliberately exact) omitted — a packed install had
dangling imports. The package now runs two single-entry builds, so each
bundle inlines its own bootstrap copy and every shipped file is
self-contained.
prepareValue admitted any cloneable value whose BOUNDED inspect rendering
fit maxValueBytes, so a huge container with a compact rendering (a
50k-element array renders as '... N more items') crossed the port raw,
bypassing the cap on both sides. The cap now measures the value's real
cross-boundary size — exact bytes for strings, the structured-clone wire
size (v8.serialize) for everything else — and oversized containers cross
as their bounded rendering instead.
ds-review-bot flagged that a typo'd or stale config key (e.g. web_fech for
web_fetch) silently applies the timeout to nothing — the tools/execute lookup
just never matches. Mirror dsh-tool-subagent's lifecycle-driven handling of a
configured-but-unregistered provider: on every tools/change (and once at load),
logger.warn each configured name still absent from ctx.tools, warning each name
at most once so a late registration silences it. Not a load-time throw — the
tool set is dynamic (cordis.yml load order, HMR), so a real tool may register
later.
Declare inject = ['tools'] since the plugin now reads ctx.tools synchronously
in apply (previously only inside event callbacks). Regenerate config-catalog
(Requires: tools) and event-producer-consumer graph.
The host's message listener trusted the compile-time WorkerToHost shape on
traffic from a peer that runs model code: postMessage(null) threw in the
listener and crashed the host process; forged log/done messages bypassed
maxLogBytes/maxValueBytes (the worker-side LogBuffer and prepareValue cap
only honest flows); and the error-reply renegotiation re-echoed a forged
non-cloneable call id, throwing outside any catch.
Every inbound message now passes a runtime shape gate that validates and
REBUILDS it field by field (junk drops without a throw; call ids must be
numbers, so replies are always clone-plain; forged extra fields never ride
along). One host-side ledger bounds everything landing in logs — honest
port entries, forged ones, and stray pipe bytes — at the single documented
maxLogBytes, with the shared in-band truncation marker emitted host-side
when the ledger trips first; the completion value is re-capped host-side
through the same prepareValue (with exactly the truncation suffix as slack
so honest worker-capped values pass unchanged), and done error text is
bounded. Also folds the stray-capture budget into that shared ledger
(round-1 finding B: it was a second maxLogBytes on top of the documented
shared cap).
The shipped backend of the code-execution seam, per the Code Mode RFC's
worker-thread section: one fresh Node worker per run, executing the
model's TypeScript after a host-side type-strip (wrapped in an
async-function shell so top-level return/await parse, sliced back out
position-preserved), bindings bridged over the message port under
hostile-peer rules (own-property name lookup, at-most-once replies,
post-settlement drops, null-prototype namespaces), logs streamed eagerly
with an in-band truncation marker, and two independent budgets — measured
event-loop busy time (computeMs) plus a never-pausing wall ceiling
(maxWallMs) — funneling into worker.terminate(). env: {} and execArgv: []
keep the isolate hermetic; disposal aborts in-flight runs and awaits
worker exits.
The worker entry loads unbuilt via Node's native type stripping
(src/worker.ts, erasable-only) and ships built as a sibling tsdown bundle
(lib/worker.js); tests/built-lib.e2e.ts pins the built load path under
plain node and joins the built-artifact smoke gate. Unit suites cover the
bootstrap in-process (fake port) and the runtime over real workers,
per-file 100%.
Codex flagged that the load-path smoke disposed the fiber only at the end,
so a leaked stale tools/execute wrapper would still pass. Add an explicit HMR
test: after fiber.dispose(), a configured tool receives the caller's own
signal unwrapped (the derived deadline is gone), matching the repo's
"dispose must reach quiescence" rule.
Model-facing tool-call budgets were tangled into each capability's schema
(bash timeoutMs, web_fetch timeout_ms) with no shared home. Add a
tools/execute around-dispatch waterfall to dsh-tools whose base next() is
the dispatch-with-normalization thunk, and a new @deepseek-ai/dsh-timeout-policy
plugin (packages/timeout/) that arms a per-tool deadline on exec.signal and
returns a structured TOOL_TIMEOUT when it wins. Migrate web_fetch (drop the
model-facing timeout_ms) and web_search onto it; the fetch provider keeps its
timeout only as a resource backstop for direct callers. bash and hook command
execution keep BASH_TIMEOUT unchanged.
Named the plugin timeout-policy (not the RFC's tool-timeout) so it does not
trip the gen-tool-catalog packages/*/tool-* completeness guard, and replace
exec.signal by in-place mutation before next() since cordis waterfall next()
ignores passed arguments. RFC moved to implemented/ recording both deviations.
A client-callback throw only becomes a JSON-RPC error RESPONSE to the
agent's session/request_permission — runScenario itself kept going, so a
tolerant agent could treat the error as a denial and the scenario would
pass, or worse, record: the impossible click baked into fixture and
golden, green on every replay. The mismatch is now captured as a harness
error while the agent is answered plain cancelled (a well-defined path
it cannot reinterpret), and the step loop rejects the run on it as soon
as the in-flight step settles. The spec asserts the rejection instead of
the agent-side error echo.
InputScript gains an optional permissionAnswers queue, consumed FIFO by
the harness's requestPermission handler. Each entry selects by option
KIND (allow_once, reject_once, …): option ids are agent-issued randoms a
committed script cannot know, while kinds are the ACP-stable vocabulary,
so the client maps kind → the offered optionId at answer time. An absent
or exhausted queue answers cancelled — existing scenarios and goldens
are untouched — and a scripted kind the request never offered throws,
surfacing as a JSON-RPC error on the permission request: the scenario
scripted an impossible click.
This is what lets an approval-flow suite (the sandbox composition) drive
allow/reject round-trips deterministically from input.json, per the
shared-acp-snapshot RFC.
A scripted fake ACP agent bin (tests/fixtures/fake-acp-agent.ts) speaks
real newline JSON-RPC through the REAL runScenario spawn path (tsx
loader, temp cwd, env plumbing); every behavior — prompt outcome,
session/new rejection, persisted logs, filesystem noise — comes from a
behavior.json beside the fixture, so specs script whole subprocess runs
from data. harness.spec.ts drives every step op, both expect-error arms,
the permission-stub default, env forwarding, workspace seeding, and the
harvest ordering/noise/fallback branches. suite.spec.ts runs the factory
for real at collection time: a replay suite over committed synthetic
fixtures and a record suite over a temp copy (write-back never touches
the committed tree; ACP_SNAPSHOT_SPEC_BOOTSTRAP=1 re-bootstraps it),
plus direct cases for the exported pure helpers. The suite factory's
pure helpers (childFixturePaths, fixtureContext, normalizedHeaders,
headerDeltaCount) are exported for those direct specs.
Two branches carry justified v8 ignores, both structurally unreachable:
the waiter in-bounds guard (noUncheckedIndexedAccess) and waitForExit's
already-exited race guard (both call sites sit one synchronous frame
after stdin.end()/kill()). The fake bin substitutes the session/new cwd,
not process.cwd(), into scripted logs — the realpath difference
(/private on darwin) is exactly what the real bin's header carries.
packages/support/acp-snapshot/src is at 100% statements, branches,
functions, and lines under the per-file gate.
The snapshot tier's machinery leaves examples/acp-agent/tests for
packages/support/acp-snapshot (@deepseek-ai/dsh-acp-snapshot), where the
coverage gate measures it and a second example can consume it instead of
forking it: harness.ts (runScenario, parameterized by an AgentUnderTest
{binScript, configPath, tsconfigPath} instead of module constants),
normalize.ts (moved verbatim), and suite.ts (defineAcpSnapshotSuite — the
per-scenario golden/log compares, record write-back, per-suite header pin
with its uniformity guard, and the fixture guard block, lifted from
acp.snapshot.ts). The example file collapses to its scenario table plus
one factory call; env reading (DSH_SNAPSHOT) stays at that edge.
The exactly-one-pin meta-test generalizes from the hardcoded text-turn
name to "exactly one per suite" — which scenario pins is the scenario
table's reviewable choice (per-suite pinning per the proposal RFC).
Extraction parity: pnpm run test:snapshot is 36 passed + fs-policy-reject
failing BEFORE AND AFTER (BSD-sed environment failure, reproduced at the
base commit in a clean worktree — the recorded golden's sed -i syntax is
GNU-only), with zero byte changes under examples/acp-agent/tests/snapshots/.
Coverage for the new src files lands in the next commit.
New group packages/code-runtime/ with the interface package
@deepseek-ai/dsh-code-runtime, per the Code Mode RFC: abstract CodeRuntime
service (run() resolves program failures as an error field, rejects only
for seam misuse), the CodeRunRequest/CodeBindingNamespace/CodeRunResult/
CodeLogEntry/CodeRunFailure vocabulary, and readonly language/isolation
backend descriptors. Registered in the tsconfig maps, packages/README,
architecture service map, and the doc-graph service-role classification;
catalogs regenerated.
The RFC's one forward path token to the worker package becomes an npm-name
mention until PR3 creates that directory (verify-package-paths is
drift-scoped: the now-existing group made the token checkable).
docs/architecture.md ceiling 1630 -> 1640: the doc gained a genuinely new
capability-service row; the row itself is already minimal.
