The executable golden proves the packaged VFS and Python SDK boundary, but it does not pin the editor-facing ACP updates, and the existing ACP scenarios exercised Code Mode, subagents, and workflows only as separate stories.
Add a dedicated both-mode replay composition that mounts a Cordis marker, inspects it through a real Code Mode worker dispatch, runs a direct spawn child and a workflow-owned spawn child, then unmounts the marker. Commit normalized ACP frames, parent and child logs, and the composition-specific request-header pin.
Use cordis_inspect as the Code Mode sub-dispatch instead of registering another dynamic tool: the executable snapshot already owns tool-set mutation and header-delta coverage, while this keeps the ACP header contract stable and focuses the new golden on transport presentation and cross-feature composition.
A preset names a bundle of the two mechanism knobs — request =
workspace-write + ask, yolo = danger-full-access + never — so the editor
shows ONE 'Permissions' select where the sandbox-mode and approval-policy
tiers stay orthogonal capabilities (the Codex /approvals shape: presets over
two dials). ctx.permission (dsh-permission) owns the config-defined table,
validates the default preset's bundle against the composed knob defaults at
load (fails loud), and writes a switch THROUGH: one log-only
permission/preset event (the audit fact reverse-mapping cannot recover —
the planned 'agent' preset shares request's knob values and differs only in
composed policy) plus each knob event via its own setter, deduped — a
net-zero switch appends nothing. Every knob consumer keeps reading its own
fold, untouched.
The current preset DERIVES from the effective knob values — the fold breaks
bundle ties, a knob state outside the table is the reserved 'custom' value
(a state, not an error: shown while it holds, switchable FROM, never a
target), and defaultPreset disappears (zero-event state reverse-maps from
the composition defaults).
The ACP bridge drops the two per-knob selects for the one preset select
(advertised only when ctx.permission is composed); pending/anchor/no-op
semantics carry over unchanged, with the no-op echo acknowledged before
vocabulary validation so a client re-pushing a derived 'custom' current
never errors. The sandbox variant example composes the
service with a workspace-write default; the permission-switching,
escalation-approved and escalation-rejected scenarios are re-recorded under
it (escalations now target an outside-workspace /tmp path under
danger-full-access, self-cleaning) and config-options is re-authored on the
single-select wire.
sandbox-acp-agent's whole surface moves into examples/acp-agent following
the example's existing multi-variant shape (the code-mode/both-mode
overlays): sandbox.cordis.yml + its replay overlay, the four snapshot
scenarios (config-options / mode-switching / escalation-approved /
escalation-rejected) as a 'sandbox' header class over per-scenario
configPath — goldens byte-identical, zero re-recording — and
escalation.e2e.ts unchanged apart from the config path. One ACP example
remains; demo:sandbox-acp keeps its name and boots the variant. References
(both RFCs, group/package READMEs, the examples table and test map, the
e2e workflow comment) now point at the variant.
Pure mechanical rename now that the package's internals are the
worker-thread engine: directory, package name, spec/e2e filenames,
module tags and logger prefixes, tsconfig/knip/run-gates/AGENTS.md
references, example cordis.yml plugin ids, doc links; catalogs
regenerated and the lockfile refreshed.
Brings in the refreshed base (master merged through the stack after #203
and #205 landed), including the acp-snapshot extraction (#204), and
re-ports this PR's snapshot-suite extensions onto the extracted package:
- dsh-acp-snapshot's Scenario gains headerClass and configPath; the suite
factory pins the request header PER CLASS (construction rejects a
missing or duplicated class pin), forwards a scenario's configPath to
the harness (RunOptions.configPath overrides AgentUnderTest.configPath),
and a new fixtures meta-test asserts every pinning fixture carries
exactly one request/header and no deltas.
- The acp-agent example's thin scenario table re-registers code-mode-turn
and both-mode-turn with their overlay configs and per-class pins; the
committed fixtures replay unchanged.
- The package's synthetic suites cover the new surface (explicit
headerClass on one suite, the default on the other, a configPath
override through the fake bin, and the two construction throws).
Three findings from the PR-4 convergence round:
(A) A root-undefined binding argument passed normalization untouched, so
the sub-call DISPATCHED and only then failed the tool/code-dispatch append
(Session.append rejects undefined event data) — a sub-call executed with
no log record, violating the nothing-executes-unlogged contract. And the
tool received the SAME object later handed to the append, so a tool
mutating its args desynced the logged record from what was dispatched (or
re-poisoned the append). jsonNormalizeArgs now rejects undefined up front
with a model-correctable message and returns TWO independent parses of the
canonical JSON text: the tool gets one, the event logs the sibling —
identical by construction, mutation-proof.
(B) The bridge built its bindings record with plain-object assignment, so
a registered tool named __proto__ hit the prototype setter and silently
vanished (the runtime host resolves binding names as own properties). The
record is now null-prototype with defineProperty, mirroring the
worker-side namespace build.
(B) The header-pin sanity assertions ran only inside NON-pinning
scenarios, so a class consisting solely of its pinning scenario (the two
Code Mode classes) would accept a re-recorded pin carrying several headers
or a header-delta. A fixtures meta-test now asserts every pinning fixture
directly.
The dsh-tools half of the Code Mode RFC (its fourth, final change): the
registry gains its first config — mode: native | code | both — and OWNS how
its tools reach the model. 'code' contributes exactly one wire tool,
run_code, plus a lazy tools:sdk prompt section declaring every other tool
as a generated TypeScript API (jsonSchemaToTs: total over the defineTool
subset, unknown degradation, lexicographic byte-identical rendering);
'both' ships both representations; 'native' is byte-for-byte the old
behavior. Non-native modes fail every assembly loudly without a
typescript-language ctx.codeRuntime.
run_code's dispatch bridge: JSON-normalizes each binding argument before
dispatch (what dispatches is what the tool/code-dispatch event logs — the
append can never fail on payload shape; BigInt/circulars reject that one
call), serializes all program tool calls through a per-run queue (even
Promise.all — no concurrency-safety metadata yet), routes every sub-call
through tools/pre-execute → tools/post-execute (a deny rejects the
program-side promise), drops sub-call additionalContext (no safe outlet
mid-run; pinned), owns a run-scoped abort that follows the outer signal in
and fires on settlement (in-flight sub-dispatch aborted, queued abandoned,
queue drained before returning), and converts a failed run into
CodeRunFailedError → a structured isError carrying kind + captured logs.
tool/code-dispatch joins SessionEventMap by declaration merging (log-only;
deriveMessages ignores it).
The composed surface: the tools config forwards through agent-core and
both app packages; examples/code-agent + demo:code run the worker runtime
under mode code (keyless boot smoke + a with-key e2e proving the collapsed
[run_code] header, the dispatch events, and the file the program wrote);
two new snapshot scenarios (code-mode-turn, both-mode-turn) record the SDK
section, collapsed header, dispatch events, and result card — each its own
header-pinning class (the harness gains per-scenario config overlays and
per-class pins). Catalogs, graphs, cookbook, hooks-bridge notes, and the
RFC (moved to implemented/, restructured to decision-era headings) updated
in the same change.
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.
Residuals from the Codex re-review:
1. A system delta's insert was flattened to one token, so deltas differing
only in inserted-line count compared equal. Now one {{system}} token per
inserted line — position AND extent survive, content does not.
2. The live uniformity guard folded only request/header snapshots, so a
mid-run header CHANGE (request/header-delta) could diverge from the pin
invisibly. Non-pinning runs now assert zero header-delta events: a
scenario that legitimately changes its header mid-run exists to show
that change, so it must pin (fail-loud until it does).
Codex review findings on the pinned-header change:
1. scrubRequestHeaders flattened a request/header-delta's whole
system/tools payload to one token, so two meaningfully different deltas
compared equal. Now the structural facts survive — keepStart/keepEnd
line positions, added/removed/changed tool NAMES — and only the bulk
(inserted prompt lines, schema bodies) is tokenized.
2. The one-pin design rested on an unasserted premise (all sessions
compose the same header). Every non-pinning scenario now asserts, live,
that each request/header its run produces equals the pinned fixture's
header (both sides normalized against their own volatile values), so a
session-dependent header fails loud until it gets its own pin.
Verified the guard bites: perturbing the pinned fixture's prompt fails
a non-pinned scenario with the intended message.
3. RFC de-slopped per docs/AGENTS.md: no PR reference, no SHOULD
spec-speak; Decision/Verification/Consequences updated for 1 and 2.
Every session.jsonl fixture embedded the full composed system prompt and
complete tool-schema list in its request/header event (~8 KB on one line,
identical across the suite), so any prompt or tool-schema edit forced a
re-record or hand-edit of every fixture — see the dynamic-workflows PR for
the churn pattern this removes.
Now exactly one scenario (text-turn, flagged pinsHeader) commits and
compares that content verbatim; every other fixture stores and compares it
as {{system}}/{{tools}} tokens via the new pure scrubRequestHeaders
normalizer (applied to both compare sides and to record-mode writes, so a
re-record cannot reintroduce the content). request/header-delta payloads
are scrubbed the same way; config/reason stay verbatim — a model swap
SHOULD churn every fixture, a prompt edit should not. Replay is unaffected:
script derivation reads only assistant/chunk events.
Fixture meta-guards enforce the split: non-pinning fixtures must be fixed
points of the scrub, the pinning fixture must not be, and exactly one
scenario pins. Committed fixtures migrated through the same function.
Docs: pinned-header RFC (implemented/testing), base snapshot RFC + testing
policy + llm-replay module doc/README updated.
A new capability family at packages/workflow/ in the bash seam shape,
modeled on Claude Code's dynamic workflows: the model writes a JavaScript
orchestration script (export const meta = {...} + plain-JS body), a runtime
executes it, and the script — not the conversation — holds the loop, the
branching, and the intermediate results.
- dsh-workflow (ctx.workflows): abstract WorkflowService + run vocabulary
(WorkflowRun whose result NEVER rejects) + observe-only workflow/* events
carrying data snapshots (id + meta, never the live run), per-listener
contained like subagent/*.
- dsh-workflow-vm: in-process node:vm engine. Meta extraction via a
string/comment-aware scanner (template interpolation rejected; literal
evaluated alone in an empty timed context; statement blanked line-
preservingly so stacks keep script line numbers). Hooks: agent(prompt,
{label, phase, schema, model}) over ctx.subagents, parallel(), pipeline()
(no cross-stage barrier), phase(), log(), args. Fatal-vs-null discipline:
hook misuse (unknown/deferred options, bad arguments, unsupported
schemas, tripped caps, seam start failures, cancellation) throws fatal
WorkflowErrors the combinators RE-THROW — never dissolved into the
per-item null reserved for child failures. Realm boundary: inbound values
materialized by descriptor walks that never invoke accessors (defineProperty
copies, __proto__-safe); outbound values rebuilt in-realm via the
context's own JSON.parse. Determinism bans (Date.now/Math.random/argless
new Date) kept so future resume support cannot break scripts. Caps and
timeouts are validated Config. Every hook promise carries a no-op
rejection consumer (app-boot exits on unhandled rejections).
- dsh-tool-workflow: the model-facing workflow tool, synchronous like
dsh-tool-subagent (start → await → try/finally dispose; abort bridged;
non-completed → isError). Generic render card titled by a textual
meta.name sniff. The tool description carries the authoring contract.
Wired into examples/{coding-agent,acp-agent} with explicit-ask-only
guidance. Coverage at every tier: unit (meta scanner, materializer incl.
counting-getter and __proto__ regressions, combinator semantics,
concurrency ceiling, caps, cancellation, no-unhandled-rejection abandon),
integration over the real spawn stack, with-key e2e (real two-phase run +
the tool through the registry pipeline), and a recorded ACP snapshot
scenario (workflow-run, 1 child session). RFC:
docs/rfc/implemented/feature/2026-07-05-dynamic-workflows.md (deferred
work explicitly listed). AGENTS.md budget 1575 → 1590 for the new group's
layout line.
Load both hook bridges in the ACP example (dsh-hooks-claude → ./hooks.json,
dsh-hooks-codex → ./codex-hooks.json) so the full-transcript snapshot tier can
exercise each dialect against the real app. An absent config file is a silent
no-op, so a scenario carries only the file it needs and the other bridge
vanishes — verified byte-identical against every pre-existing snapshot.
Add a scenario per hook point × its headline Decision outcome, both dialects:
UserPromptSubmit block (authored, keyless) + context-fold, PreToolUse deny/ask,
PostToolUse block/context, Stop force-continue. The mid-turn scenarios are
recorded against the real API with the hook active, so the model's reaction to
a denied/blocked/force-continued turn is part of the replayed transcript.
SessionStart and SubagentStart are deliberately excluded (detached best-effort
inject races the log position — a recorded golden fails 10/10 on its own
replay), as is SubagentStop (observe-only, zero transcript footprint — a golden
could never be proven to fail). Both stay on the bridges' unit coverage. See
docs/rfc/implemented/testing/2026-07-04-hook-snapshot-matrix.md.
Replace the "bag of optional fields" tool-presentation types
(ToolCallPresentation / ToolResultPresentation / ToolTerminal) with a
card-tagged discriminated union — the standing FIXME(tool-presentation).
A tool declares one render intent per call/result and the ACP bridge
switches on `card`:
ToolCallView = generic | terminal | diff
ToolResultView = generic | terminal
The `diff` card is new: fs write/edit now emit an ACP {type:'diff'}
content block (an editor's inline diff), which the old shapes could not
express. The bridge also relativizes a file card's title against the
session cwd (mirroring claude-agent-acp's toDisplayPath) while keeping
locations/diff paths raw, and derives the no-capability fenced console
fallback from a terminal result's output. read gains the window-in-title
(`Read foo.txt (5 - 8)`) and an always-set location line, matching the
reference adapter field-for-field.
Migrates all three producer families (tool-fs, tool-bash, tool-todo) and
the sole consumer (the ACP bridge) together — the source does not compile
piecewise. Adds snapshot coverage for the terminal _meta path (a new
capability-advertising scenario) and re-records the fs goldens to show the
diff cards. Applied-hunk (result-time, context-line) diffs need a new
result/event shape and are a follow-up.
RFC: docs/rfc/implemented/architecture/2026-07-02-tool-render-intent-union.md
The fs-policy gate throws FS_NOT_OBSERVED when the model edits a file it
never read; that rejection surfaces as a failed tool_call_update, but no
snapshot pinned it — a regression that dropped or mis-rendered the failed
card would pass every gate. Record a scenario that edits a seeded file
without a preceding read: the edit is vetoed, the file stays unchanged on
disk, and the transcript shows the pending edit card followed by a
status:'failed' update carrying the policy error.
Five recorded ACP snapshot scenarios exercising read/write/edit end-to-end
through the real acp-agent subprocess, replayed keyless in CI:
- fs-read — read a seeded file (read tool + presentation + observed-state)
- fs-write — create a file (write, no prior version guard)
- fs-edit — read then literal-replace (read-before-edit authorization)
- fs-write-overwrite — read then rewrite (replaceIfVersion after a read)
- fs-read-window — read lines 5-8 with offset/limit (windowing + the offset
surfaced as the tool_call location line)
The goldens confirm the tools render with their new presentation — Read/Write/
Edit <path> titles, read/edit kinds, and `locations` (fs-read-window carries
`{path, line:5}`) — and that the prompts steered the model to the fs tools, not
bash (zero bash calls in any golden). Recorded against the real API, filtered to
the new scenarios so no existing fixture churned.
The two bridge plugins that run a user's existing Claude Code / Codex hook
config on the harness's typed interception seams, built on the shared
dsh-hook-protocol library. A bridge is a faithfulness adapter, not a power
tool: anything it does a native cordis plugin does more powerfully — the
bridge exists only to run UNMODIFIED external hooks.
- dsh-hooks-claude: CC dialect. Seven hook points (SessionStart,
UserPromptSubmit, PreToolUse, PostToolUse, Stop, SubagentStart,
SubagentStop), CC per-event stdin payloads, env + ${CLAUDE_PLUGIN_ROOT}/
${CLAUDE_PROJECT_DIR} substitution, literal-or-regex matcher.
- dsh-hooks-codex: Codex dialect — a deliberate subset. Five hook points,
always-regex matcher, snake_case payloads (turn_id/model, no trailing
newline), no env/substitution, block-only decisions.
Both map the neutral merged outcome onto the seam's typed Decision and stamp
an explicit {kind:'plugin'} source on injected context (so it is never
mislabeled as a user prompt). Config parse-failure is contained; only command
hooks run. updatedInput is logged+warned (input rewrite deferred); the Stop
loop-guard is deferred (TODO).
Tests: per-file 100% — config-parse unit branches + per-seam mappings
end-to-end through the REAL loop + REAL bash + REAL shell scripts (scripted
mock model only) + a real-Loader export-shape guard. A keyless ACP snapshot
scenario (hook-prompt-block) proves a UserPromptSubmit hook blocks a prompt
end-to-end (rejected turn -> ACP cancelled, hook/* events in the log); a
with-key e2e (hooks.e2e.ts) proves a PreToolUse hook blocks real bash
(verified on disk). The snapshot normalizer now scrubs hook/result.durationMs.
RFC: docs/rfc/implemented/feature/2026-06-30-hook-bridges.md
Record the `todo-plan` snapshot scenario: a real prompt drives the model to call
todo_write, and the golden captures the resulting `plan` sessionUpdate (three
entries, priority synthesized as medium, status 1:1) plus the persisted
todo/write event. Registered in SCENARIOS; replays deterministically keyless.
Add a with-key coding-agent e2e that verifies the WORLD — a real model call to
todo_write lands a todo/write event whose snapshot is a valid, one-in-progress
list — not the agent's self-report. Wire tool-todo into the e2e harness.
The seed-boundary change made fork-child replay route correctly but shipped
with no recorded fork scenario — the seedLength slice was exercised only by
llm-replay unit tests and a persistence round-trip, never by the full-transcript
snapshot tier. Add two recorded scenarios that drive a real fork child through
it:
- subagent-fork: parent completes a turn, then forks one child (child fixture
carries a non-zero seedLength, the boundary the replay slice consumes).
- subagent-mixed: parent completes a turn, then delegates once via spawn
(seedLength 0) and once via fork (non-zero seedLength) in one transcript —
the first scenario to drive two subagent backends at once, exercising both
branches of the slice.
Both need a completed turn-1 so the fork seed is a non-empty completed-turn
prefix (a turn-1 fork seeds empty = spawn, which would not exercise the slice).
Removing the slice turns both scenarios red (the fork child receives the
parent's recorded chunks), proving the guard bites.
ACP (out-of-process) subagent replay remains a different shape, still tracked
as TODO(acp-subagent-replay).
The snapshot tier was built single-session: dsh-llm-replay served calls from
one global positional cursor, and the harness harvested one session log. A
subagent runs as a second agent with its own session, so a parent→child
scenario could neither replay deterministically nor harvest the child's log.
This resolves the TODO(subagent-snapshots) deferral from the subagent RFC.
- Stamp the calling session id onto the model request: GenerateOptions.sessionId
(typed Branded<'SessionId'> to avoid the dsh-llm↔dsh-session cycle), set by the
agent loop from agent.session.id. Adapters ignore it; an llm/stream listener
routes by it.
- Key replay per session: dsh-llm-replay loads the parent log plus one per child
(childFiles / $DSH_SNAPSHOT_CHILD_FILES), derives a script per recorded session,
and binds each live (freshly-random) session to a recorded script by first-call
order — parent first (earliest createdAt, first to stream). Keys by WHO calls,
so it survives a future concurrent/backgrounded subagent; a global cursor would
not. An unrecorded extra session fails loud.
- Harvest every log: the harness collects all .jsonl across cwd buckets, ordered
primary-first (top-level, then children by createdAt), and RunResult exposes the
plural sessionLogs. The spec writes each back on record (session.jsonl +
session.<n>.jsonl) and diffs each against its fixture on replay.
- Wire the subagent seam + spawn + fork + tool into the acp-agent example (both
cordis configs) and add two nested scenarios recorded against the real API:
subagent-spawn (parent + 1 child) and subagent-multi (parent + 2 children, 3
sessions). Both replay keyless in the default gate.
A new RFC documents the design (docs/rfc/implemented/testing/). Single-session
replay is unchanged (a call with no sessionId is one anonymous primary session).
TODO follow-up: a dedicated branded-ids package could own the SessionId brand and
dissolve the cross-package cycle note; out of scope for this testing PR.
The per-kind fixture guard claimed no-model scenarios need no session.jsonl, but
the replay path requires one for ALL scenarios: runScenario() passes
`fixtureFile: <dir>/session.jsonl` unconditionally and llm-replay's
loadReplayScript() throws "fixture not found" when it is absent and no override
replaces it. A no-model scenario ships a header-only session.jsonl that derives
to an empty script. The guard + its comment now match that reality, so a future
no-model scenario following them won't fail at subprocess startup.
Model-driving ACP snapshot scenarios shipped both session.jsonl (the
replay fixture) and session.golden.jsonl (the expected re-persisted log).
For recorded scenarios the normalized fixture and golden were byte-identical
— pure duplication. Remove session.golden.jsonl entirely: every model
scenario now has at most one committed session-log artifact, session.jsonl,
which doubles as the replay source AND the expected produced log.
The snapshot test compares the replay run's persisted log against the
session.jsonl fixture, normalizing BOTH sides — but each against its OWN
volatile values, not a shared context. A raw harvested fixture bakes in the
recording run's session id / cwd / timestamps, distinct from the live replay
run's; since normalizeSessionLog scrubs cwd by exact string match, the
fixture must be normalized against its own header (new fixtureContext helper)
or its stale recorded cwd would leak unscrubbed and the compare would fail.
The session side uses a normalized-string toEqual, NOT toMatchFileSnapshot,
so a run never overwrites the fixture.
Authored override scenarios (error-finish, cancel) now hold their expected
produced log in session.jsonl. Verified llm-replay ignores the fixture for
model chunks when an override exists: loadReplayScript() returns the override
array and never reads config.file, so committing the full expected log there
does not affect replay behavior.
The required-fixture guard is now per-kind: every scenario needs input.json +
stdout.golden.jsonl; model scenarios need session.jsonl; authored ones
additionally need replay.override.json. Updates the ACP-snapshot-tests RFC to
the reduced fixture set and moves the proposing RFC proposed -> implemented.
Master's "fix(acp): align prompt and workspace contracts" made session/new
reject a non-empty additionalDirectories / mcpServers (widening the workspace
scope is unimplemented). Add a `reject-extra-dirs` scenario + a
`newSessionExpectError` input op that pins this editor-facing contract: the
bridge answers with `-32602 Invalid params: additionalDirectories is not
supported`. Keyless, deterministic, no model call.
(session/load replay — the other new master behavior — needs a two-phase
seed-then-load harness and is left for a focused follow-up.)
Tianyi Cui