write/edit failures with FS_STALE_VERSION or FS_NOT_OBSERVED now reach the
model with the correct recovery instruction appended (re-read / read, then
retry) while preserving the structured code and chaining the cause. The
edit-intent waterfall sits inside the same try, so the policy's
FS_NOT_OBSERVED refusal is remediated too. Re-recorded the fs-policy-reject
keyless snapshot and the bilingual README pairs.
An empty read window (byte cap below the first selected line: `lines: []`
with `totalLines > 0`) dropped `offset` from the persisted presentation
meta, so a replayed read card could not report where the window starts or
where a continuation resumes. Carry `offset` on `FsReadMeta`,
`ReadResultView`, and the `presentationMeta` projection, and validate it in
`readMetaFromMeta` (1-based integer; the first line number may not fall
below it). Re-record the ACP fixtures and the cordis api catalog.
Also correct the Note's `parallel-file-reads` golden path
(examples/tui-agent -> apps/cli) and record the pre-card replay-degradation
tradeoff in the Decision section.
A filename whose extension is an Object.prototype key (foo.constructor,
foo.__proto__) resolved to the inherited member through the plain-object index,
so a function reached the read card's lang hint and failed the tool-output JSON
validation, failing an otherwise successful read. Look the extension up as an
own property only. Added rejection tests, converted the zh Note headings to the
all-English sibling convention, and named the parallel-file-reads terminal
golden as the TUI-unchanged evidence in the Testing section (both languages).
readMetaFromMeta narrows the opaque persisted meta boundary, so beyond
shape it now rejects replayed JSON that is well-typed but semantically
invalid: line numbers must be 1-based integers, totalLines a non-negative
integer, and line numbers must strictly increase without exceeding
totalLines. Any violation declines to the generic fallback.
Sync the public ToolResultView contract across the core/tools and
tool-fs READMEs and docs/core-data-structures/tools for the fourth
result-view member and the ReadFileLine vocabulary, and expand the
Agent Note Testing section with the new rejection paths and the snapshot
evidence this PR carries.
The read tool's result carries structured numbered lines, but only the
model-facing envelope text reached the client. Add a card:'read' result view
(ReadResultView) projecting {path, lines, totalLines, lang} through the tool's
output.presentationMeta so presentResult reproduces it on live and replay
paths; the pending call stays a generic read card. A UI without the read
capability falls back to the envelope-stripped content, so the TUI is
unchanged. The web consumer that renders the line-numbered view is a follow-up.
- fs-sandbox: delegate the mutation with the freshly re-canonicalized target
(not the stale one), so the checked identity is the mutated identity — a
symlink swapped in after resolve() can no longer escape workspace-write.
- tool-fs: map a denial to an FsError carrying FS_SANDBOX_DENIED (not a plain
Error), so ToolRegistry keeps the structured code on result.error for
retry/observers while the message stays the shared marker.
- sandbox-local: derive the Seatbelt writable set from the shared
writableRoots() helper, so the profile and the fs fence cannot drift.
- gen-doc-graphs: ctx.sandboxPolicy is owned by dsh-sandbox-policy and read only
by the sandboxed executor/provider (the tool layers use the pure fold).
- docs: bash-sandbox/bash/permission READMEs and bash.md reflect the relocated
policy home and the sandbox/mode rename; drop the stale stdout.golden.jsonl.
Extend SandboxMode enforcement from bash to the filesystem tools, the sandbox
RFC's deferred cross-family phase.
- dsh-sandbox-policy (new, ctx.sandboxPolicy): the single home for the
deployment default mode + workspaceRoot and the per-session override event,
renamed bash/sandbox-mode -> sandbox/mode and moved here with its fold/setter.
Decouples the bash seam from dsh-session.
- dsh-fs-sandbox (new): SandboxedFileSystem extends LocalFileSystem and fences
write/edit by the per-call mode (read-only denies, workspace-write contains to
the workspace + temp roots via the shared writableRoots, danger passes
through); reads pass through. Structured FS_SANDBOX_DENIED; in-lock parent
re-canonicalization. A policy fence in trusted code, not a kernel boundary.
- dsh-sandbox: the shared escalation kit (writableRoots, the strictly-wider
ladder, denial/hint markers, approveEscalation) both tool families use;
approveEscalation takes a structural approver so dsh-sandbox gains no
approval/agent dependency, and both tools stay duplication-free.
- tool-fs: write/edit advertise sandbox_permissions/justification under a
confining ctx.fs, map FS_SANDBOX_DENIED to the shared [sandbox: ...] marker,
and resolve the same one-approved-wider retry.
- examples/acp-agent: composes sandbox-policy + fs-sandbox, drops the gating
that disabled the fs stack under confined modes.
RFC docs/rfc/implemented/feature/2026-07-14-cross-family-fs-sandbox.md; the old
sandbox RFC's In-process/deferred/FAQ sections updated to shipped fact.
The read tool's isConcurrencySafe rationale called the fs/observed recorder
"commutative" and said concurrent reads "converge to one observed version",
overstating the guarantee: the WeakMap record is last-writer-wins. Safety comes
from write/edit re-checking the version in their in-lock CAS (a stale
observation only forces a later edit to fail closed with FS_STALE_VERSION), as
the RFC already states. Align the read comment, the ToolDefinition JSDoc, both
READMEs, and the regenerated catalogs.
New doc-sync gate verify-export-jsdoc walks every module-level exported
name under packages/*/*/src and requires description prose everywhere,
plus @param per parameter and @returns on non-void annotated returns for
function-like exports, public class methods, properties, and accessors.
The parsing + check helpers move out of gen-cordis-catalog.ts into a
shared scripts/jsdoc.ts so 'documented' means one thing on both gated
surfaces.
Deliberate exemptions (documented in the RFC): heritage-declared class
members (the seam declaration is the doc's one home — the one checker
query in an otherwise pure-AST walk), cordis plugin-protocol slots
(name/inject/reusable/Config/apply, top-level and static), constructors,
overload implementations, declare-module augmentation bodies, and
re-export statements (checked at the defining module).
The 203 under-documented exports the gate found at adoption are filled
in this change, so the gate lands green; generated catalogs/graphs are
regenerated for the shifted line pointers.
RFC: docs/rfc/implemented/process/2026-07-06-export-surface-jsdoc-gate.md
One principle: every fact in the assembled prompt has exactly one owner.
- dsh-system-prompt: merge-extensible AssembleContext on assemble();
a variable(name, provider) registry; {{name}} interpolation in
renderPrompt, strict (unknown/valueless/malformed references throw);
duplicate section and variable names rejected; assembly carries
resolved section text + variables through the assemble waterfall.
- dsh-agent declares AssembleContext.agent; dsh-agent-loop registers
the agent:persona section (order 0 - identity renders before tool
guidance) and the model/cwd variables, and drops its string join:
renderPrompt(assembly) IS the full prompt.
- Tool guidance moves to its owners: descriptions carry per-tool
semantics; sections only cross-call habits (tool:bash exit-code
habit at order 105; read's not-shell nudge). todo/subagent need no
section - their descriptions already carry the contract.
- SubagentProvider.inheritsParentContext (spawn/acp false, fork true);
dsh-tool-subagent derives truthful per-provider wording and resolves
the provider at load (backend must be listed first).
- Example personas shrink to identity + behavior with {{model}} (and
{{cwd}} in the ACP tree); the welcome banner stops enumerating tools.
RFC: docs/rfc/implemented/architecture/2026-07-05-prompt-variables-and-tool-guidance-ownership.md
A Codex review pass on the draft caught four real gaps and two solid
suggestions; all addressed except one pushed back on the merits:
- hooks-claude/hooks-codex: stderrSummaryMaxChars was the one new knob
with NO range validation — a negative/NaN cap would silently
misbehave inside slice(). Both bridges now assert a positive integer
at the TOP of apply() (before the config-file parse's early return,
so a bad value fails the load loudly), with rejection tests.
- tool-fs: the read caps count lines/chars/bytes, so positive-FINITE
was too loose (a fractional readLimit would flow into windowing
arithmetic and the schema description). All four now require a
positive integer, matching tool-web's cap.
- Doc drift the gates cannot catch: tool-web's README tools table
still named WEB_SEARCH_MAX_RESULTS as the mechanism; compact-basic's
README/module doc and the compaction-capability-seam RFC still
described estimation as fixed char/4 rather than the charsPerToken
default.
- subagent-acp: the dispose graces were tested only at the
startAcpRun level, so a regression that stopped threading plugin
config into AcpRunSpec would have survived. A provider-path test now
drives the trap-escalation scenario through ctx.subagents.start with
small config graces and bounds dispose at 4s.
Pushed back on: converting compact-basic's charsPerToken to a
schemastery field. The package's whole config is deliberately
hand-rolled (resolveConfig, every threshold REQUIRED with no default —
a documented design posture); one schemastery field beside it would be
incoherent. The knob is cordis.yml-reachable, defaulted, and validated,
which is what the convention requires; migrating the package to
schemastery wholesale is pre-existing config-surface hygiene out of
this change's scope.
_Kerman
Chinesezjc
Turtle
Tianyi Cui
kingwl
Dudu-0223
Yichen Jiang