Review follow-up: the blanket 10s default correctly targets worker-thread
cold-start races (starting, first-script-line, async child-registration
messages — genuinely CPU-bound under CI contention), but four waits assert
something different — that the HOST reacted PROMPTLY to an event that
already happened (a settled result, an observed worker death). Those had no
cold-start left to wait on, so the generous default just widened the window
a real regression could hide in.
Verified by injecting a 6s delay into the settle-reap's abort call: the
un-overridden helper's test still passed in ~6s. The same mutation now fails
in ~1s with the explicit 1000ms override restored on all four sites (the
abort-on-settle test's own assertion, the two worker-death cancel/dispose
reap checks, and the dispose-ack-race check). The other 12 waits keep the
10s default — they run BEFORE a result is awaited, waiting on the worker to
actually start rather than on a host reaction. Doc comment corrected to
describe the split instead of claiming every wait is a cold-start race.
Review finding, valid: under the registry's Code Mode the assembly's
only wire tool is run_code, which the plan allowlist filtered out —
leaving the model with NO tools at all, the exit review included. The
composition exists today (the acp-agent example ships a code-mode
overlay), so plan mode bricked it outright.
run_code is a transport, not a capability: every bridged sub-call is
serialized back through ToolRegistry.execute() carrying the same agent,
so tools/pre-execute judges each capability individually — exactly like
native calls. Both layers now exempt it by name: the filter keeps it
visible (tests pin plan-mode Code Mode assembly = ['run_code']) and the
gate passes the wrapper while the same run's write sub-call still
denies with the plan-mode reason.
Documented residual, same class as the prepend-after-load one: the SDK
section renders from the registry's store, so a plan-mode program may
be offered bindings whose dispatch the gate then denies — nothing runs
that a native call could not.
Review finding at the seam of two surfaces this branch added to the
same stdin: with an ask_user_question (or plan-review) prompt active,
the line handler dispatched every line as the answer first, so
'/mode plan' typed mid-question was recorded as free-text feedback —
model-visible in the tool result — and the mode never changed. Command
handling now runs before answer dispatch: the command executes, the
question stays pending and still owns the next non-command line. A
literal '/mode…' free-text answer is the trade-off deliberately spent —
a swallowed command that becomes review feedback costs far more than
that contrived answer shape.
Review finding with a real in-repo instance: the structured runtime's
per-spawn final-assembly wrapper (prepend, post-next) re-injects
structured_output OUTSIDE the mode filter, so a structured child in
plan mode would see a tool the gate then denies — the soft policy and
the hard gate telling different stories. The suggested fix (make the
mode filter outermost) cannot beat that instance: prepend unshifts, so
the per-spawn listener always registers later and wraps outer.
Two-part resolution instead. Semantically, structured_output enters the
shipped plan allowlist — it is a child's pure result channel, the same
ask/report class as ask_user_question and exit_plan_mode, so the
filter, the re-injection, and the gate now agree wherever a structured
child runs in plan mode. Mechanically, the filter registers with
prepend anyway: it now wraps outside every append-registered listener
regardless of load order (regression test pins a pre-registered
post-next mutator being filtered), narrowing the documented cosmetic
residual to prepend-after-load listeners only, where the gate still
covers execution. Severity note: no execution breach existed — the gate
held throughout; this closes the prompt-honesty gap.
The spec's 16 vi.waitFor sites used the 1s default timeout to wait for
worker-thread startup and child registration — CPU-bound work that blows
past 1s on a contended runner. The CI coverage lane (4 vitest workers plus
suites that spawn real subprocesses) hit this 3 times across 4 recent PR
runs, each a different subset of the cancellation/worker-death tests, each
green on rerun.
Every wait now goes through a shared helper with a 10s bound, and the file
sets a 30s test timeout to make room for it. The one deliberately tight
wait keeps its 800ms bound through the helper's override — it proves the
host (not the wedged worker's later loop turn) delivered the cancel, so a
generous bound would erase what it tests. No behavior under test changed.
spec.onError is a caller-supplied callback boundary, but the flattening
catch invoked it unguarded: a throwing sink rejected the whole async result,
breaking the seam's "result never rejects on a child-level failure"
contract (and docs/defensive-patterns.md's contain-callback-exceptions
rule). The sink's own throw is now swallowed — named as such — while the
original child failure still settles as stopReason 'error'. Regression test
drives a spawn failure through a throwing sink and asserts result resolves.
Same defect as the codex backend's, fixed there on PR #240; this is the
symmetric fix for the already-merged ACP backend.
Review hardening (the finding's ordering premise did not hold — see the
PR thread — but its failure-path kernel did): onBoundary cleared the
pending intent BEFORE appending the mode/set, so a backend rejecting
that one write lost the switch forever while the picker kept showing it
optimistically. The intent is now cleared only after the append lands;
a failed flush stays parked and the next healthy boundary converges the
log with the picker. The containment test extends to pin the re-park
and the retry.
The bridge's re-notify keeps deriving from the logged event's value —
now documented in place: the service holds ONE coalesced pending slot
(every flush reads the latest selection, so a stale flush cannot
exist), and for any other writer the logged value is the truth the
picker should track, in log order.
The three wedged-child scenarios busy-spin their worker for 1.5s while the
host waits for the start RPC; under a loaded 2-core CI runner (this branch
adds several parallel suites) the RPC lands after vi.waitFor default 1s,
failing the coverage lane three runs in a row at the same three sites. An
explicit 10s waitFor timeout (well inside each test own 15s budget) makes
the assertions load-tolerant without weakening them. Belongs upstream with
dsh-workflow-workerthread; carried here because it gates this PR.
The three-entry cordis.yml (dsh-sandbox-local + dsh-bash-sandbox at a
read-only default + dsh-approval) served over ACP: the first live approval
composition. Recorded snapshot scenarios pin the wire end to end —
config-options advertisement, the mode-switching arc as the suite pinned
header (both switches, the prompt-section delta, one changed-by-the-user
notice per knob, a confined write landing under the switched mode), and
both escalation branches over scripted permissionAnswers (a grant runs
confined under workspace-write; a rejection executes nothing and pins the
fail-closed text). The with-key escalation e2e drives a real model +
real runner + the real bridge answerer, world-verified; ci.yml snapshot
lane and e2e.yml install bubblewrap so the confined replays actually
execute.
Both RFCs move to implemented/ (Decision/Consequences form, deferred
phases tracked in their own sections), with every cross-reference flipped.
effective(session) = findLast(the session own knob events)?.value ?? the
composition-config default. One log-only event per knob, owned by its
domain (bash/sandbox-mode in dsh-bash, approval/policy in dsh-approval),
each exporting the same three-piece kit: the event declaration, a pure
fold, and THE write path — a switch IS its event; no owner service, no
facts map. Restart immunity and multi-session isolation fall out of the
log replay by construction.
Execution follows the fold on both sides: the bash tool stamps
escalation grant > session override > executor default, and the approval
seam prepends the never-gate that auto-rejects before any interactive
answerer. Visibility is two layers per knob: a per-agent prompt section
states the effective value on every request (logged through
request/header*, so what-the-model-was-told replays from the log), and an
agent/pre-step narrator injects at most one coalesced delta notice with
positional attribution (user switch vs operator/config drift). The ACP
bridge advertises one capability-gated select per composable knob with
currentValue folded per session, validates set_config_option against the
closed vocabularies, and anchors idle switches at the next turn
prompt-submit under the turn-enclosure contract.
The tool gate advertises sandbox_permissions (an enum of exactly the modes
STRICTLY WIDER than the mounted executor default — the schema makes a
non-widening request inexpressible) plus a required justification, exactly
when ctx.bash.sandboxMode reports a confining mode at registration:
composition truth, never a dead lever. An escalating call resolves
ctx.approval BEFORE anything executes with the audit-self-contained reason
"escalate sandbox to <mode>: <justification>"; allowed-once stamps the
granted mode onto that one bash request (the seam-level per-call override),
while rejected / cancelled / unavailable and the no-service / no-agent
paths each fail closed with their own error text and execute nothing. The
description teaches the flow only when the fields exist: retry the SAME
command once after a real denial, never preemptively; a rejected
escalation is final. No new session events: the attempt is an ordinary
tool/call, the decision is the approval audit pair, the outcome an
ordinary tool/result whose facts name the mode it ran under.
dsh-bash grows the per-call policy carrier: BashExecRequest.sandboxMode
(request-optional, spec required-but-nullable — the owner pattern; resolve()
is the one explicit defaulting step) and the BashExecutor.sandboxMode
capability fact (undefined in the base class — composition truth the tool
layer can read). dsh-bash-local carries the field verbatim and confines
nothing.
dsh-bash-sandbox extends LocalBashExecutor and hands ctx.sandbox the exact
argv it is about to spawn. A denial is a RESULT FACT (the command RAN;
result.sandbox.denied is orthogonal to exitCode/signal), classified
conservatively against the wrap own dialect; a RUNNER failure outranks
denial — foreground re-throws the structured SANDBOX_UNAVAILABLE, a settled
background task stamps sandbox.runnerFailed — so a broken sandbox never
reads as a failing command and the command never runs unconfined.
dsh-tool-bash renders the markers and teaches the model not to retry around
a policy denial; escalation and per-session switching are staged follow-ups.
ctx.sandbox (dsh-sandbox): confine(argv, policy) returns the argv to spawn
instead — wrapped so the process and its children run confined — plus the
enforcement completeness and the backend denial/runner-failure dialects;
no usable backend throws the fail-closed SANDBOX_UNAVAILABLE. Policy rides
per call. dsh-sandbox-local selects by platform and caches the verdict:
multi-candidate chains probe FUNCTIONALLY in preference order (Linux:
bwrap → the registry-installed node-addon-landlock-run launcher), a sole
candidate is selected unprobed (darwin: sandbox-exec/Seatbelt) and fails
closed at execution via runnerFailureSignatures; win32 is a reserved empty
chain. Profile parity is honest per backend (documented temp-area and ABI
differences; enforcement full|partial is a structured result fact).
CI: the sandbox-e2e matrix proves real-kernel confinement per rung (bwrap,
Landlock per architecture through the registry-installed launcher,
Seatbelt), failing on a silent all-skip; the packed-install rehearsal
installs the launcher family from the registry and asserts the binary
executable apart from kernel enforcement.
The ACP bridge registers the first real approval answerer: an ask for an
agent it owns becomes session/request_permission attached to the already-
streamed tool call (one-shot allow_once/reject_once only), outcomes map
conservatively (unknown optionId never grants, client cancel → cancelled),
and foreign or call-less requests delegate down the waterfall. The snapshot
harness accepts scripted permissionAnswers (FIFO; an unscripted prompt
answers cancelled, fail closed) so recorded scenarios can drive the wire
keylessly.
ctx.approval (dsh-approval): request() dispatches the approval/request
waterfall and always resolves a closed outcome — allowed-once / rejected /
cancelled / unavailable — never rejects; zero listeners fall through to
fail-closed unavailable; abort settles cancelled and discards late answers;
throwing or rogue answerers are contained as unavailable; every ask lands
the log-only approval/asked / approval/decided audit pair. dsh-tools routes
a pre-execute ask through the seam opportunistically (ctx.get) with three
distinct deny reasons, keeping the historical ask→deny degrade when the
seam is absent.
The per-session policy tier, the ACP bridge answerer, and the sandbox
escalation asker are staged follow-ups of the approval-seam RFC.
Live failure: deepseek-v4-flash answered a greeting entirely in the
reasoning channel — no text block, no tool calls. The serializer's
null-content fallback produced an assistant message with neither
content nor tool_calls, which the API 400s ('Invalid assistant message:
content or tool_calls must be set'). Because that message sits durably
in the session log, every later turn of the session re-derived the same
history and failed identically — one all-reasoning response bricked the
session permanently (log: turns 2 and 3 failing byte-identically).
content is now always the flattened text ('' when there is none); the
passback rule still keeps reasoning_content off plain turns. The old
null shape was pinned by a test whose comment claimed the wire accepts
it — live-falsified, updated together with the code, plus a regression
test for the reasoning-only shape. Existing bricked logs resume cleanly
under the fix (the poisoned message now serializes as '').
Live-session feedback (a real Zed elicitation round-trip): the model
presented its finished plan as a plain reply and asked the USER to
switch modes — the exact reversal the roadmap warns about — because the
shipped section's 'present it with the exit_plan_mode tool' read as a
suggestion. The section now says a finished plan is delivered by
calling exit_plan_mode, preferred over pasting it as a plain reply or
asking the user to switch modes — firmer, without imperatives.
ask_user_question enters the shipped plan allowlist (asking is
read-only-safe), and the section points a blocked decision at it. The
plan-acp-agent example composes the bash family (default mode only —
plan's allowlist keeps excluding it, so the two modes now demo a real
difference) plus tool-ask-user; both recorded scenarios re-recorded:
the pin now shows plan = [ask_user_question, exit_plan_mode, read,
todo_write] and post-exit default = the full eight-tool surface.
Review finding: the exit tool's direct mode/set append flipped the
folded mode while the loop could still execute further tool calls from
the SAME assistant response — a same-batch exit_plan_mode + write pair
would sail past tools/pre-execute under 'default' even though the
request was assembled under the plan-shaped header. That broke the
design's own invariant (a step's executions run under the mode its
assembly folded), which the pending-intent flush was built to hold for
user flips.
The tool now records the switch as a pending intent like every other
writer, flushed at this step's end (still in-turn); pending intents
carry a narrate flag so the exit's flush stays silent — the tool result
is its narration — while user flips keep the coalesced boundary notice.
The gate, folding the logged mode only, now provably covers the whole
batch: regression test pins approve-then-write-in-the-same-batch as
denied, and the widened toolset still arrives on the next step.
The recorded scenarios are re-recorded: the fixture now shows mode/set
landing after step/end, before the widened fallback header.
The with-key recording session the RFC deferred. plan-mode is the 'plan'
header class's pinning scenario and necessarily carries BOTH header
shapes verbatim: the plan-shaped initial snapshot and the widened
fallback snapshot the approved exit produces mid-turn — so the suite
factory's pin guards relax from exactly-one to at-least-one header (the
classmates' uniformity anchor is the pin's FIRST header; transition
headers are legal only in the pin, matching the sandbox stack's
precedent). plan-mode-reject pins the keep-planning branch: the
corrective isError carries the reviewer's free-text feedback verbatim
and the session stays in plan mode, one header, uniform with the pin.
Recording notes, encoded in the prompts: the model is pinned to RELATIVE
paths (a recorded absolute temp path neither replays on another host nor
normalizes — the normalizers scrub the run's own cwd, not the
recording's), and the recorded model never calls a filtered tool, so the
gate's deny path stays pinned at the unit tier — that refusal is the
behavior the soft layer exists to produce. The gen-tool-catalog
meta-test pins exit_plan_mode in the harvested schema list.
The proposal survives contact with the code with three amendments, per
the RFCs-are-proposals rule. (1) A mode transition logs a
request/header-delta only when expressible: adding exit_plan_mode
resorts the canonical tool list, and a pure reordering has no delta
form, so entering plan mode logs the full fallback snapshot — the
attributability claim holds either way. (2) The proposed/ skeleton
converts to the implemented grammar: Proposal → Decision, the roadmap's
staging (now history) drops to the standing Deferred list, and
Acceptance criteria + Risks fold into Consequences (what holds, then the
accepted costs, including the ACP v2 mode-removal migration). (3) The
two recorded scenarios stay pending a with-key session, recorded in
Deferred.
The docs tail completes: the cookbook's plan-mode row upgrades from
sketch to the shipped package, architecture.md gains the ctx.modes
capability row (ceiling 1640 → 1650: a new capability service's table
row does not fit the old budget), and every reference repoints to
implemented/.
Plan mode's stage 2 (RFC 2026-07-07-plan-mode). The exit tool: one
required plan argument (the durable log artifact), execute re-checks the
folded mode, then conducts the review over the user-interaction seam —
one single-select question (Approve / Keep planning) with free text open
— so an approval appends mode/set back to default in-turn and every
other outcome (keep-planning feedback verbatim, aborted, no provider)
returns the corrective isError with the mode unchanged. presentCall is a
generic card titled by the plan's first heading carrying the plan
markdown; over ACP the review rides the ask_user elicitation flow, in
the terminal the stdio prompt queue — no approval-seam dependency.
The ACP bridge maps the picker 1:1 onto ctx.modes (opportunistic, a
type-only peer edge): session/new + session/load advertise
availableModes/currentModeId, session/set_mode validates through set()
and echoes an optimistic current_mode_update (the pending mode IS the
selection; the logged mode/set lands at the boundary and, matching, is
not re-sent), and a session/event listener re-notifies on each logged
flip that differs from the last sent — the tool-driven exit updates the
picker. The feature matrix rows move from 'not modeled' to the
picker-to-modes / knobs-to-config-options division, with the ACP v2
removal direction recorded as a mechanical-migration risk.
The snapshot harness gains the setMode/setModeExpectError ops and a
scripted elicitationAnswers FIFO (cancel on exhaustion; a stray choice
string reaches the agent verbatim as a non-consenting custom answer, so
a scenario bug fails safe). The suite factory's header-pin requirement
now applies only to model-turn scenarios — a protocol-only suite has no
header content to anchor. examples/plan-acp-agent is the live
composition; its keyless modes-advertise scenario pins the wire surface
(advertisement, both set_mode round-trips, unknown-id rejection). The
recorded plan-mode approve/reject arc awaits a with-key recording
session; its texts are pinned at the unit tier meanwhile.
examples/AGENTS.md ceiling 653 → 680: the new example's required smoke
row does not fit the old budget.
Plan mode's stage 1 (RFC 2026-07-07-plan-mode): a new packages/mode/ group
with one product package owning the mode/set SessionEventMap vocabulary
(log-only, non-surface, whole-value replace), the pure foldMode, and the
ctx.modes service (list/get/set). User flips are pending intents flushed
at turn/start / step/end — turn enclosure makes an idle append illegal —
with one coalesced context/message notice when the flushed mode differs
from what the last logged request header told the model; a folded mode
the config no longer defines reads as default plus one boundary notice.
Enforcement is two covering layers: a system-prompt/assemble wrapper
filters the RETURNED assembly's tools to the mode's allowlist (and shows
exit_plan_mode IFF the folded mode is plan) beside the mode:policy
section at order 50, and a tools/pre-execute gate denies deny-by-default
against the same allowlist, judging by the logged mode only. The default
mode is the absence of policy — assemblies stay byte-identical to a
no-dsh-mode deployment.
AgentOptions.mode (declaration-merged) seeds a child's initial mode
through the same flush on agent/created; the stdio app gains /mode
(print/switch, never sent to the model) over an opportunistic
ctx.get('modes'). Config is an explicit resolve step: the built-in plan
definition (read-only allowlist; bash/subagent excluded until the
sandbox family lands) merges unless overridden, 'default' as a key
throws at load, unknown names throw at set() time.
The empty-env hardening wiped the one variable the UNBUILT worker's
loader depends on: tsx finds a tsconfig by searching up from the
worker's cwd, and a parent running outside the repo (the ACP snapshot
harness pins the repo tsconfig through TSX_TSCONFIG_PATH exactly
because its child cwd is a temp dir) lost the dsh-* paths map - the
worker then resolved workspace imports to unbuilt lib/ bundles and died
on CI with ERR_MODULE_NOT_FOUND (green locally only because stale built
lib/ masked the wrong resolution).
Forward exactly that variable when the parent carries it - loader
plumbing, not a secret; the built shape stays fully empty - and pin the
whole contract with an escape-based test: the worker env is exactly
{TSX_TSCONFIG_PATH}, the credential canary still never crosses.
Adopts #211 (Code Mode tools: run_code + the code/both-mode snapshot
scenarios). Tool-catalog expectations take the union (run_code +
task_*); the two new pinsHeader fixtures (code-mode-turn,
both-mode-turn) were recorded on master without the task runtime, so
they are re-pinned KEYLESSLY by replaying their recorded chunks against
the merged tree (same procedure as text-turn) — the fixture diff is
exactly the header delta: task tool schemas, the tool:tasks prompt
section, and the bash background wording.
Master advanced 11 commits mid-merge (the Code Mode registry integration:
mode config, run_code, the tools:sdk section, the ACP run_code cards and
unified demos). The fusion makes Code Mode scope-aware end to end:
- wireSchemas(scope): the mode-aware wire contribution is computed from the
CALLING SCOPE's visible set (scoped tools join, shadowing and restrictions
apply) and feeds the {schemas, knownNames} provider protocol.
- knownNames under the mode collapse: a per-scope RESTRICTION is runtime
state, so the universe stays pre-restriction (a restricted-away tool in
toolOrder is a normal absence) — but the MODE collapse is deployment
config, so under mode 'code' the universe is [run_code] and a toolOrder
naming a native tool fails every assembly loud (master's tested decision,
kept).
- The tools:sdk section renders per assembly CONTEXT: the SDK declares
exactly the calling agent's callable set, using the section-text provider
signature this branch already had.
- run_code bindings enumerate schemas(exec.agent) — a program can bind
exactly what its prompt promised; sub-dispatches already threaded
exec.agent through registry.execute, so scoped resolution and carriers
flow unchanged.
- dsh-tools declares both sides' new deps (dsh-scope + dsh-session);
lockfile and all generated catalogs/graphs/api-catalog regenerated.
Gates green on the merged tree: typecheck, lint, per-file 100% coverage
(2710 tests), snapshots (41), doc-sync, module graph, build, hygiene, demo
smoke.
The documented vm escape reaches process, and the worker inherited the
harness's env - so a buggy or prompt-injected script could read and
exfiltrate ambient credentials (DEEPSEEK_API_KEY et al.) without
touching a single file (ds-review-bot finding on #233).
Spawn with env: {} and a hermetic execArgv on both runtime shapes, the
same stance as dsh-code-runtime-worker and stronger than the scrubbed
env the defensive-patterns rule requires for spawned commands (a shell
needs PATH; this worker needs nothing). Ambient-channel hardening only:
an escapee keeps the process-wide privileges the trust premise already
admits - the genuine sandbox remains an engine swap.
enableRunInBackground: false removed the parameter from the advertised
schema only — the arg validator deliberately allows undeclared keys, so
a caller (or a model that has seen the parameter elsewhere) could still
force run_in_background: true and start background work past the
deployment's opt-out, in both tool-bash and tool-subagent. Both
producers now refuse the forced key loud in execute(); tests pin the
refusal (and that nothing spawns) alongside the untouched foreground
path; the schema-omission-is-advertising rule is recorded in the
runtime RFC and both READMEs.
Master brought 50 commits (the tool-cordis group, dsh-code-runtime + worker,
the tools/execute around-dispatch seam + timeout-policy, repeat-tool-guard,
agent/session-prefix, the ui reorganization). Beyond the ten textual
conflicts, the merge reconciles master's new seams with this branch's
scoped-registration world:
- tools/execute (new waterfall around core dispatch): dispatched with the
SAME exec.agent carrier as the pre/post waterfalls — an agent.ctx wrapper
times/retries only its own agent's calls — and its base thunk resolves the
tool through the caller's visible view (get(exec.name, exec.agent)), so a
scoped/shadowed tool dispatches and a restricted-away global stays
UNKNOWN_TOOL. Declared this: Scoped<ToolRegistry> with the scope-filtered
doc sentence; invariants table + verify-scoped-dispatch pin it (21 events).
- agent/session-prefix (new waterfall, once per loop instance): composed via
the fused agentEvents dispatcher (scope-filtered like every agent-subject
event), declared this: Scoped<Agent>, table-pinned. agent/pre-step keeps
master's new sessionPrefix parameter with this branch's Scoped this.
- timeout-policy reads the budget through the caller's visible view
(get(exec.name, exec.agent)): a scoped tool's own timeoutMs governs its
calls; a global name-twin's budget is never misapplied to a shadowing
per-agent variant.
- tool-cordis: cordis_inspect's tools section lists the CALLING agent's view
(its description promises "what you can call"); the sandbox tool façade's
reads resolve through the mount's own scope, mirroring where its register
lands writes; sandboxRegisterTool's return type carries the exact-disposer
union honestly. dsh-scope declared as peer+dev with the project reference.
- doc-sync chain unions master's verify-cordis-api with this branch's
verify-scoped-dispatch; the generated catalogs, event matrix (the
zero-dispatcher guard passes over master's new events), module graph, and
the cordis api-catalog are regenerated on the merged surface.
Full gate sequence green on the merged tree: typecheck, lint, per-file 100%
coverage (2668 tests), snapshots (38), doc-sync, module graph, build,
hygiene, demo smoke.
pku-xht
kingwl
Yichen Jiang
Hypatia May
Dudu-0223
pku-xht
Tianyi Cui
imccyu