124 Commits

Author	SHA1	Message	Date
Tom Boucher	03738824de	enhance(#2586): stop installing Codex context-monitor hooks without metrics (#4367)	2026-09-06 05:46:49 -04:00
github-actions[bot]	b3906c66f6	chore: sync next package version to 1.13.0	2026-09-06 02:10:29 +00:00
Behruz Nassre Esfahani	5ad9a36f35	fix(#4255): resolve reviewer-lane effort from the lane, not from gsd-plan-checker (#4275) ... `review-lane plan` resolved every cross-AI reviewer lane's reasoning effort by spawning `query resolve-execution gsd-plan-checker --host <slug>`. The agent id was a hardcoded literal, so `--host` chose only the argv RENDERING while the LEVEL always came from the installed plan-checker's frontmatter — `low` under every shipped model profile. Every prompt-fed lane therefore ran at a fast structural verifier's effort, and because the rendered argument is a CLI config override it silently beat the effort the operator had configured for that CLI. At `low` a large source-grounded prompt makes a model end its turn with no final message, so the lane came back empty and its stub read as a crash. Effort is a property of the review, so the lane declares it. Two new fields on ReviewerLane — `effortConfigKey` (`review.effort.<slug>`) and `defaultEffort` — carried through each capability manifest and the generated registry, set on the three lanes with an argv effort channel and null on the other nine. A new pure `resolveLaneEffort()` resolves config key -> lane default -> nothing, where "nothing" emits no effort argument at all and the reviewer CLI's own configuration decides; `inherit` selects that path explicitly and an unrecognized level falls back to the lane default rather than being forwarded to a CLI that would reject it. The host's negotiated effortSurface still gates the rendering, so ADR-1239/#2481's trust boundary holds on this path too. Resolving in-process also removes up to twelve subprocess spawns per review. The empty-output stub now names the effort the lane ran at and distinguishes a clean exit from a timeout kill, a non-zero exit, and a process that never ran — `status` is null for both a timeout and a signal, so those were indistinguishable before. The hint is hedged: a clean empty exit is most often a model stopping short, but it is also consistent with a CLI writing its output elsewhere. Also: the capability validator now knows both fields, rejects a malformed key or an out-of-vocabulary default, and rejects a default declared without a config key (a level the operator could never override). An existing end-to-end row in tests/effort-surface-axis.test.cjs asserted the old coupling; it now configures the lane's own key and pins the decoupling in the same real spawn, with the agent execution tier set to a level that must not appear. Emitted-Drift-Ack-Growth: review.md — the effort/model resolution-order table this fix adds. The workflow is where an operator looks to find out which knob set a lane's model and effort; leaving the new key undocumented there is the same invisibility that made the plan-checker coupling survive this long. Emitted-Drift-Ack-Growth: review.md — the effort/model resolution-order table this fix adds. The workflow is where an operator looks to find out which knob set a lane's model and effort, so leaving the new key undocumented there is the same invisibility that let the plan-checker coupling survive. Claude-Session: https://claude.ai/code/session_01CRMEuzNMWn3gs5uUW2ghcF Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com> Co-authored-by: Tom Boucher <trekkie@nomorestars.com>	2026-09-05 05:25:44 -04:00
Dennis Alexis Valin Dittrich	925a363879	enhance(#4032): apply configured agent tool grants (#4238) ... * test(4032): add failing installed-agent grants contract Cover global and project agent_tools precedence at the real Claude installer seam before adding implementation. * feat(4032): apply configured agent tool grants during staging Resolve selector-level global and project config once per staging call, then append validated grants before runtime conversion. * test(4032): cover host grant and quoted MCP contracts Exercise installed host artifacts and prove ZCode must treat quoted MCP scalars like plain MCP grants. * feat(4032): apply configured agent tool grants across runtimes Move augmentation and scalar identity into the converter seam so every staged artifact preserves host policy. * fix(4032): register agent tool grants in configuration Accept documented agent_tools config without unknown-key warnings.\n\nKeep installer fixtures on the shared temporary-directory helper. * fix(4032): translate configured MCP grants for Kilo Reuse the converter-owned scalar decoder so quoted canonical grants reach Kilo's native permission keys without altering other host policies. * fix(4032): decode YAML-escaped tool grants * fix(4032): emit valid inline agent tool grants * fix(4032): reject invalid trailing-colon grants * test(#4032): cover cross-review remediation gaps * fix(#4032): close cross-runtime grant gaps * test(#4032): expose Kimi global project context * fix(#4032): preserve Kimi project config context * chore(#4032): add release note * test(#4032): expose fork review regressions * fix(#4032): address fork review findings * test(#4032): make byte-stability assertion portable Compare repeat installs at one root so platform-specific path rendering cannot masquerade as an agent_tools behavior change. * chore(#4032): bind changeset to upstream PR 4238 * fix(#4032): address trek-e review findings (2,3,4,5,6,7,8) Fixes fail-closed decode-failure handling in ZCode's mcp__ stripper, a comment-only `tools:` header mis-parse that silently dropped configured grants, and a naive comma-split that could tear a quoted scalar containing a literal comma. Documents Kilo's inherent `{server}_{tool}` MCP-permission-key collision (external, fixed format — not ours to widen) and locks the existing first-seen-wins resolution in with a regression test. Opts kimi/kimi-code out of the ADR-1235 pre-converter path-rewrite step: routing Kimi through that pipeline (needed so project-scoped agent_tools selectors reach it) was short-circuiting Kimi's own neutralizeKimiAgentPrompt, which expects the original ~/.claude/gsd-core text rather than a pre-rewritten Kimi path. Extends the fast-check token pool and per-runtime install coverage with the missing comment/comma/broad-runtime cases the prior review flagged as untested. * docs(#4032): add CONTEXT.md glossary entries for agent_tools resolver + pre-converter step Documents readGsdEffectiveAgentTools (Install Model Override Resolver Module) and the appendAgentTools pre-converter pipeline step (Runtime Artifact Conversion Module), per contributor-standards.md's new-seam glossary requirement (finding 1). * fix(#4032): address agy adversarial review findings An agy (gemini-3.8-flash-high) adversarial pass over the prior review-fix commit found the fixes for findings 3, 4, 6 and 8 had unfixed sibling gaps, plus a genuine new regression and two CONTEXT.md inaccuracies: - ZCode's comment-only `tools: # note` header matched the inline-value branch instead of falling through to the block-list scan, so a following mcp__* item leaked through unstripped — the exact defect finding 4 fixed in appendAgentTools, unfixed in this sibling function. - Reverted capabilities/kimi-code/capability.json's noPathRewrite: true. kimi-code uses the standard 'agents' kind with converter: null (not kimi-agents — confirmed by reading the descriptor, not its prose description), so it never went through the pipeline change finding 5 fixed, and disabling its path rewrite broke every ~/.claude/ embed in its shipped agents instead. - decodeToolScalar never stripped a trailing ` # comment` from a bare (unquoted) scalar, so a comment after a block-list item, or after an appended grant on an inline line, became part of the "tool name" — fixed at the source (one call site fixes every consumer). - appendAgentTools's comment-index scan wasn't quote-aware, so a `#` inside a quoted scalar (`"mcp__server #1"`) was mistaken for a comment start and corrupted the quote. - parseFrontmatterTools (Kimi/Qwen's tool-list reader, downstream of appendAgentTools's own output) had the same naive comma-split and comment-only-header gaps as findings 4 and 6, unpatched. - The all-runtime smoke test's presence assertion was built on a guessed omit-list; empirically only 7 of 17 runtimes keep an arbitrary mcp__ grant recognizable, replaced with a verified allowlist. - CONTEXT.md claimed a `project:<agent>` selector prefix that does not exist (project override is a same-key merge across two config files) and mislabeled stageAgentsForRuntimeWithConverter's module. * fix(#4032): address full-PR review (Opus critical/ponytail + agy) A whole-PR pass (critical-code-reviewer + ponytail-review on Opus, plus a second agy full-source adversarial pass) surfaced defects the earlier finding-scoped passes couldn't reach: - appendAgentTools corrupted a `tools:` line whose ENTIRE value is a leading quoted scalar (`tools: "Read"` -> `tools: "Read", Write`, invalid YAML) — there is no safe line-surgical rewrite here, so it now refuses to touch that shape instead of emitting broken frontmatter. - decodeToolScalar's malformed-trailing-quote check ran BEFORE comment stripping, so a bare tool name with a quote inside its own trailing comment (`Bash # note: "internal"`) was wrongly rejected. Reordered. - findUnquotedCommentIndex (added in the prior remediation commit) was built on a wrong model of YAML: a `#` after whitespace starts a real comment in a plain scalar regardless of nearby quote characters — verified against the actual parser. The one case that DOES need protection (a leading quoted scalar) is now refused outright above, so the quote-tracking scan was dead weight solving a problem that no longer reaches it. Removed; reverted to the plain `[ \t]#` scan. - Kilo has a SEPARATE agent-frontmatter parser (convertClaudeToKiloFrontmatter, distinct from the buildKiloAgentPermissionBlock fixed earlier) with the same comment-only-header and naive-comma-split gaps as findings 4 and 6 — unfixed in both its src/ and bin/install.js copies. Fixed in both, exporting splitToolScalars for bin/install.js to reuse rather than reimplementing it. - Pipeline docstring in stageAgentsForRuntimeWithConverter still listed 5 steps, omitting appendAgentTools (now step 3 of 6). - docs/CONFIGURATION.md didn't state that a --global install still discovers agent_tools from the cwd's .planning/config.json (confirmed intentional and already covered by a dedicated test, not a bug). - Removed install-engine.cts's deps.cwd injection seam: zero callers or tests ever populated it. Two claims from this round were verified and rejected, not fixed: prototype pollution via a `__proto__` selector key (empirically confirmed `Object.prototype` is never touched — only reassigns the resolver's own local object's prototype, with no observable effect), and a `*` grant value crashing YAML parsing as an alias reference (empirically confirmed it parses as plain scalar text, no crash). A pre-existing, unrelated defect (extractFrontmatterField returns null for block-list `tools:` on Copilot/Antigravity/Cursor/Codex/Qwen, affecting two shipped agents today) was filed as a follow-up rather than fixed here — it predates #4032 and isn't caused or worsened by this PR. * fix(#4032): update stale slug-derivation-drift-guard fixture line normalizeKimiSkillName's real closing brace moved from line 616 to 635 as a side effect of this PR's edits to runtime-artifact-conversion.cts; the MAJOR-1 fixture's hardcoded realEndLine had gone stale. * fix(#4032): address CodeRabbit findings on projectDir threading and flow-sequence tools bin/install.js's installAgentsKindStandalone call site omitted the projectDir argument the function already supports, so a global install through this legacy branch silently fell back to the runtime config dir instead of process.cwd() when resolving project-scoped agent_tools grants — inconsistent with the sibling installOpencodeFamilyArtifacts call site, which already threads it correctly. appendAgentTools' leading-quoted-scalar bailout did not cover a YAML flow sequence (`tools: [Bash, Read]`): splitToolScalars tore it apart on the in-sequence commas and appended past its closing bracket, producing invalid frontmatter. Extended the bailout regex to also refuse a value starting with `[`, matching the same "whole node, nothing may follow" reasoning already applied to quoted scalars. --------- Co-authored-by: CI Rebase Check <ci@gsd-redux> Co-authored-by: Test <test@test.com> Co-authored-by: Tom Boucher <trekkie@nomorestars.com>	2026-09-05 04:52:45 -04:00
Dennis Alexis Valin Dittrich	a262ad6b61	fix(#4148): dispatch wave-pre step hooks (#4185) ... * fix(#4148): dispatch wave-pre step hooks External capabilities can render step hooks before a wave, but the execute workflow consumed only contributions and silently skipped every step. Reuse the shared dispatch contract before executor spawning and pin the capability-validator boundary with a red-first regression. Emitted-Drift-Ack-Growth: execute-phase.md — wave-pre now carries the missing generic step-dispatch contract before executor spawning * test(#4148): pin wave-pre dispatch ordering * test(#4148): pin wave-pre dispatch contract * chore(#4148): bind upstream changeset PR * chore(#4148): restore fork changeset identity * fix(#4148): align wave-pre dispatch contract Mirror the sibling wave-post all-shapes clarification while pruning redundant prose so the rebased workflow remains below its frozen byte ceiling. Emitted-Drift-Ack-Growth: execute-phase.md — wave-pre now carries the missing generic step-dispatch contract before executor spawning * chore(#4148): restore upstream changeset identity * fix(#4148): align wave-pre capability guidance * docs(#4148): identify wave-pre manifest input Name the third-party manifest trust origin at the wave-pre dispatch boundary so the reviewer-requested validation guidance matches wave-post. Emitted-Drift-Ack-Growth: execute-phase.md — wave-pre now carries the missing generic step-dispatch contract before executor spawning * docs(#4148): preserve execute-phase byte budget Remove a redundant advisory label while retaining the non-blocking contract, keeping the reviewer-required trust-boundary wording at the enforced 93,400-byte ceiling. * fix(#4148): mark wave-pre manifest-input validation as security-relevant Reviewer nit on PR #4185: wave-pre's step-dispatch sentence had the (third-party manifest input) parenthetical but dropped the ⚠ marker that wave-post's parallel sentence (execute-phase.md:1044) carries, losing the visual flag that this validation is security-motivated. Trims the redundant "of one" from "not one shape of one" to reclaim the 4 bytes the marker adds — the ADR-857 byte-margin gate (tests/claude-orchestration.test.cjs) leaves zero slack at the 93,400-byte ceiling. * fix(#4148): trim wave-pre step-dispatch prose to clear ADR-857 byte ceiling Merging next's unrelated growth (#3990's TDD_APPLICABLE conditional) pushed execute-phase.md 116 bytes past the 93,400-byte ceiling, failing CI on all three platforms. The security-relevant ⚠ marker and ref.command validation call-out (added per prior reviewer nit) are preserved verbatim per the pinned regression test in capability-registry.test.cjs; only the non-pinned connective prose is trimmed. * fix(#4148): recalibrate execute-phase.md self-imposed margin, restore security marker next grew execute-phase.md by ~230 bytes across two unrelated merges during this fix (#3990's TDD_APPLICABLE conditional, then a further step-extraction commit), consuming this test's own self-imposed 93,400 safety buffer under ADR-857's actual, unmodified 93,600 ceiling (docs/adr/857-capability-system.md:22). The wave-pre step-dispatch sentence cannot shrink further without dropping one of the pinned substrings this same test file asserts on (kind=="step", loop-hook-dispatch, never blocks or redirects executor spawning, Validate `ref.command`). Raises the self-imposed margin to 93,550 (still 50 bytes under the real, untouched ADR ceiling) and restores the ⚠ marker the prior reviewer round required for the ref.command validation call-out, which byte pressure had dropped. * fix(#4148): restore full ref.command validation wording, drop self-imposed margin Adversarial review (agy/gemini-3.8-flash-high) flagged two issues in the prior CI-recovery commit: 1. Trimming "in-context before any shell use" from the step-dispatch warning weakened the inline operational instruction (the reader is told WHAT to validate but not the specific in-context-not-shell mechanism the referenced loop-hook-dispatch.md:45-51 threat model requires). Restored it - the merge with next since the last commit freed enough real margin (77 bytes under the untouched 93,600 ADR-857 ceiling) to afford it without any margin change. 2. The prior commit self-imposed margin bump (93400 to 93550) was, on reflection, the wrong lever: it is a number this PR invented, not an ADR value, and re-bumping it every time next grows execute-phase.md is a losing pattern (already needed twice in one session). Removed the redundant assertion; the same line existing bytes-under-93600 check against the real, frozen ADR-857 ceiling (docs/adr/857-capability-system.md:22) is the actual invariant and is untouched. workflow-size-budget.test.cjs tier hard cap (98304 bytes, extract-not-bump by design) remains the correct backstop for runaway growth. --------- Co-authored-by: CI Rebase Check <ci@gsd-redux> Co-authored-by: Test <test@test.com> Co-authored-by: Tom Boucher <trekkie@nomorestars.com>	2026-09-05 02:56:37 -04:00
Tom Boucher	acb903c2e8	enhance(#3661): make the code-review hook point configurable (#4159) ... * feat(#3661): make the code-review hook point configurable Add `workflow.code_review_point` (`execute:post` default, or `execute:wave:post`) so a multi-wave phase can run code review once per wave instead of once at the end, scoped to what changed since the phase's prior review. The code-review capability now declares its step at both loop points via a new generic `pointFrom` step field: `pointFrom` names an enum config key, and the step is only active at its own `point` when that key resolves to a matching value. `_resolvePointGate` (capability-activation.cts) is the single shared implementation consumed identically by loop-resolver.cts and capability-state.cts, and capability-validator.cjs enforces that `pointFrom` references an enum key whose values cover the declaring step's own point. code-review.md's manual-invocation gate now reads `workflow.code_review` directly instead of probing registry presence at the hardcoded execute:post point (so manual `/gsd-code-review` keeps working regardless of which automatic point is configured), and its file-scope tiers narrow to what changed since the phase's last review commit when one exists. execute-phase.md's wave-post step dispatch gets a small, precedented carve-out so the code-review skill still receives its required phase argument when dispatched generically (caught by the isolated spec review). Closes #3661 Emitted-Drift-Ack-Growth: code-review.md — #3661 adds a point-aware config gate check and LAST_REVIEW_COMMIT-based incremental scoping to the file-scope tiers. Emitted-Drift-Ack-Growth: execute-phase.md — #3661 adds one carve-out sentence so the wave-post generic step dispatch passes PHASE_NUMBER to the code-review skill. * docs: backfill changeset PR number for #3661 (#4159) * fix: scope tests/io.test.cjs's fs.writeSync fault-injection mocks by fd Five fault-injection mocks in the "bug #1008" describe blocks intercepted every fs.writeSync call regardless of file descriptor, and several threw or truncated unconditionally on the first call. This surfaced as an intermittent macOS CI failure: node:test's own IPC channel back to the parent process (which also goes through fs.writeSync internally) could get a bogus injected error or truncated write if node's internal machinery called it while one of these mocks was active, corrupting the message frame the parent tried to deserialize ("Unable to deserialize cloned data.", location tests/io.test.cjs:1:1, uncaughtException — a whole-file IPC crash, not a test assertion failure). Root cause confirmed by a working counter-example already in the same file: the "#3912 A6" mocks gate on `fd !== 2` before any fault injection and were never implicated. Applied the same fd-scoped pattern to the five unscoped mocks (four output()-targeting tests gate on fd 1, one error()-targeting test gates on fd 2), and added a regression test proving an unrelated fd passes through untouched while the fault-injection mock is active. Found while verifying #3661; unrelated to that change's own diff. --------- Co-authored-by: sim <sim@local>	2026-09-02 11:01:54 -04:00
Tom Boucher	ff0361071d	feat(#3653): add review.models.cursor — wire modelArg/modelConfigKey for the cursor reviewer lane (#4160) ... * feat(#3653): add review.models.cursor — wire modelArg/modelConfigKey for the cursor reviewer lane cursor-agent exposes --model (204 selectable models) but the cursor lane declared modelArg: null / modelConfigKey: null, so review.models.cursor was rejected as an unknown config key and the #1517 reviewer-instances escape hatch silently discarded a configured model at modelExpansion. Wire the lane the same way codex already is: inject {{model}} into args right after -p, set modelArg to --model, and declare modelConfigKey as review.models.cursor plus its config schema entry. An unconfigured lane still invokes byte-identically to today. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * docs(#3653): add changeset for review.models.cursor Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(#3653): update co-change surfaces that assumed cursor has no model key gsd-test surfaced three surfaces still hardcoding "cursor declares no modelConfigKey", broken by wiring review.models.cursor: - tests/reviewer-config-federation.test.cjs: the #3691-narrows-#2797 invariant test listed cursor among lanes that must own no model key. - tests/settings-integrations.test.cjs: the #3651 keyless-lane test listed cursor as keyless, including a live config-set assertion that now correctly succeeds instead of failing (swapped to qwen). - gsd-core/workflows/settings-integrations.md: the settable-keys enumeration and two prose call-outs still named cursor as keyless. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * chore(#3653): acknowledge deliberate growth of settings-integrations.md settings-integrations.md grew 4 bytes because it now enumerates review.models.cursor as a settable key alongside the other reviewer lanes, matching the modelConfigKey wired for cursor in this PR. Emitted-Drift-Ack-Growth: settings-integrations.md — adds review.models.cursor to the settable-keys enumeration and removes cursor from the two keyless-lane call-outs, matching #3653's modelConfigKey change Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * chore(#3653): fix malformed Emitted-Drift-Ack-Growth trailer The previous commit's trailer was separated from Co-Authored-By by a blank line, splitting it into an earlier, non-trailer paragraph — git's trailer parser only recognizes the last contiguous block. Restating it here immediately adjacent to Co-Authored-By so both parse as trailers. Emitted-Drift-Ack-Growth: settings-integrations.md — adds review.models.cursor to the settable-keys enumeration and removes cursor from the two keyless-lane call-outs, matching #3653's modelConfigKey change Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * docs(#3653): backfill changeset pr number pr:0 -> pr:4160 now that https://github.com/open-gsd/gsd-core/pull/4160 exists. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> --------- Co-authored-by: sim <sim@local> Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com>	2026-09-01 21:40:20 -04:00
Tom Boucher	4dfc46bbe7	enhance(#3348): add a context-drift pre-check gate to plan-phase (#4147) ... * test(#3348): add failing-first coverage for the context-drift gate * feat(#3348): add context-drift pre-check gate for plan-phase Compares each phase's *-RESEARCH.md/*-PATTERNS.md/*-VALIDATION.md/*-SPEC.md effective last-changed time (git commit time, falling back to mtime for uncommitted edits) against *-CONTEXT.md's, so plan-phase no longer silently reuses an upstream artifact that predates a decision added to CONTEXT.md after that artifact was derived from it. Deterministic, no model call. New `gsd_run verify context-drift <phase>` command, sibling to the existing verify.codebase-drift/verify.schema-drift gates in the drift capability. Warn-only by default (workflow.context_drift_precheck), with an opt-in workflow.context_drift_action: block escape hatch. Wired at plan:pre in plan-phase.md, before both the RESEARCH.md and PATTERNS.md reuse decisions. * fix(#3348): address code-review findings — raw-text-match, stale comment, import placement, duplicated phase resolution * fix(#3859): pin the real commit's diff.ignoreSubmodules to match the empty-diff probe The #3859 empty-diff guard decides whether a submodule bump would land using `--ignore-submodules=dirty`, overriding the caller's `diff.ignoreSubmodules` config. The real `git commit -- <paths>` that follows was never given the same override, so under a bare `diff.ignoreSubmodules=all` repo config the two calculations disagree: driven on git 2.39.5 (Debian bookworm, the linux-node24 test-matrix image), the guard correctly stands aside but the scoped commit itself then silently fails (exit 1, no error text) for a gitlink bump it had just confirmed would be recorded, surfacing as commit_failed instead of committed:true. Pin `-c diff.ignoreSubmodules=dirty` onto the scoped commit call too, so the probe and the commit it protects can never diverge. Harmless when no submodule path is involved (driven: identical outcome on an ordinary scoped file, with and without the flag). * fix(#3348): guard resolvePhaseDirByToken's exact-match fallback against path traversal * fix(#3348): retarget phase-enumeration-drift exemption to the consolidated resolvePhaseDirByToken helper cmdVerifySchemaDrift's inline readdirSync was already function-scoped-exempt in lint-phase-enumeration-drift.cjs as a single-phase LOOKUP (not a current-milestone enumeration). This PR's refactor pass lifted that block into a shared helper, resolvePhaseDirByToken, also used by the new cmdVerifyContextDrift — the guard tracks exemptions by enclosing function name, so the readdirSync now lives in an unexempted function and started firing. Move the exemption to resolvePhaseDirByToken (same written reason, now covering both callers) instead of migrating to listAllPhaseDirs, which would introduce two real behavior deltas here: it catches readdirSync failures internally (old code let them throw) and sorts results by phase number before matchPhaseDirs picks matches[0] (old code used raw, OS-dependent readdirSync order). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(#3348): satisfy lint:ci — slash form, capability registry regen - docs/features/context-drift-gate.md used the deprecated /gsd: colon form; docs are never passed through the install-time slash-form converters, so lint-docs-command-form requires the hyphen form. Regenerated docs/FEATURES.md from the corrected fragment. - Regenerated gsd-core/bin/lib/capability-registry.cjs after editing capabilities/drift/capability.json (lint:generated-sync). * fix(#3859): pin the real commit's diff.ignoreSubmodules via env, not argv -c The prior fix pinned `-c diff.ignoreSubmodules=dirty` onto the scoped commit's argv via `commitArgs.unshift(...)`. `-c key=val` must precede the `commit` subcommand, so this shifted `commitArgs[0]` from `'commit'` to `'-c'` for every scoped commit call, breaking 17 position-based assertions in the commit-files pathspec regression suite that read `a[0] === 'commit'` to find the commit invocation among recorded git calls. `execGit` already accepts an `env` option merged onto `process.env` before spawning. Git honors `GIT_CONFIG_COUNT`/`GIT_CONFIG_KEY_0`/`GIT_CONFIG_VALUE_0` as a per-invocation config override functionally identical to `-c key=val`, expressed via env instead of argv. Passing that env alongside the existing commitArgs (still `['commit', ..., '--', ...stagedPaths]`, argv unchanged) fixes the real commit's effective diff.ignoreSubmodules to match the empty-diff guard's probe without moving anything in argv position 0. Scoped to exactly the canScope branch, matching the probe's own preconditions and leaving no behavior change for commits the probe never evaluated. No test file changes needed — the 17 previously-failing assertions test argv[0] against the array passed into execGit, which never changes. * fix(#3348): register verify-context-drift in the check subcommand router The drift capability's new plan:pre gate declares check.query "verify.context-drift", which normalizes to `check verify-context-drift`, but no such subcommand was routed — phase6-capstone-conformance's uniform-block-field test failed with "Unknown check subcommand" for every declared gate query. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(#3348): extend #1592's exact-key-list snapshot for the new context-drift config keys tests/capability-registry.test.cjs asserted an exact, hardcoded snapshot of the drift capability's config keys. #3348 legitimately adds two new keys (workflow.context_drift_precheck, workflow.context_drift_action) for its own plan:pre context-drift gate — extend the expected set (and clarify the assertion message) without weakening the test's exactness. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(#3348): reconcile E2's exemption-migration pin with the resolvePhaseDirByToken extraction #3348 (an earlier commit on this branch, e4b80ad81) extracted cmdVerifySchemaDrift's inline phasesDir readdirSync/matchPhaseDirs block into the shared resolvePhaseDirByToken helper (also used by the new cmdVerifyContextDrift), and retargeted lint-phase-enumeration-drift.cjs's function-scoped exemption from cmdVerifySchemaDrift to resolvePhaseDirByToken accordingly — cmdVerifySchemaDrift no longer contains a line the guard's detectors match, so it needs no exemption. tests/phase-locator.test.cjs's E2 test still pinned the exemption to the old name (cmdVerifySchemaDrift), unaware of the migration. Update E2 to match the same "migrated call site's exemption must move, not duplicate" pattern the test already applies to cmdRoadmapAnalyze and cmdInitMilestoneOp just below it: drop cmdVerifySchemaDrift from the still-exempt list and add symmetric assertions that it no longer carries the exemption while resolvePhaseDirByToken now does. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(#3348): fix two self-contradicting/nondeterministic tests in context-drift.test.cjs 'always exits 0 (query command contract)' included the no-phase-arg case, which contradicts the file's own earlier 'errors with usage message on missing phase arg' test (that case legitimately exits 1 via the Usage error) — drop it from the always-exits-0 cases. 'degrades to mtime comparison outside a git repo' and '...in a repo with no commits' relied on real wall-clock ordering between two back-to-back writeFileSync calls to prove CONTEXT.md is newer than RESEARCH.md; on a fast filesystem both can land in the same mtime tick, producing a tie that computeContextDrift's strict `<` correctly treats as not-stale, so stale_artifacts comes back empty. Make both tests deterministic via explicit fs.utimesSync instead of relying on timing (CONTRIBUTING.md: never assert elapsed wall-clock time). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(#3348): add context_drift_precheck:false to the plan:pre all-off fixture The "all plan:pre when-keys false" fixture explicitly disables every known workflow.* plan:pre toggle, but didn't yet know about the new workflow.context_drift_precheck key (defaults to true), so the new drift context-drift gate stayed active and broke the empty-activeHooks assertion. Emitted-Drift-Ack-Growth: plan-phase.md — adds the #3348 context-drift plan:pre pre-check section (new ## 4.6); this PR's own diff, not incidental drift. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * docs(#3348): backfill changeset PR number (pr:0 -> 4147) --------- Co-authored-by: sim <sim@local> Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com>	2026-09-01 21:39:37 -04:00
Tom Boucher	c0fd2e3f4c	feat(#3673): add dispatch.maxConcurrency axis and dispatch-capacity query (#4162) ... * test(#3673): add failing tests for dispatch.maxConcurrency axis and dispatch-capacity CLI route Extends tests/host-integration.test.cjs with negotiateHostCapabilities maxConcurrency negotiation coverage (test matrix rows 1-15, including a fast-check property test) and a new #3673 dispatch-capacity CLI route describe block spawning the real gsd-tools.cjs (rows 16-25). Extends tests/host-integration-validator-parity.test.cjs with an all-19-descriptor maxConcurrency presence/validity sweep (row 26) and adds a hostile-input validator test to host-integration.test.cjs (row 27). The dispatch.maxConcurrency field does not exist yet, so these tests fail. * feat(#3673): add dispatch.maxConcurrency axis, negotiation, validator parity, and the dispatch-capacity query Adds a numeric dispatch.maxConcurrency sub-field to the Host-Integration Interface (ADR-1239 Phase 1), following the existing dispatch.isolation sub-field pattern: DispatchCapability interface, SAFE_DEFAULTS/PROFILE_BASELINES floors, and a negotiateHostCapabilities branch that passes through a positive safe integer and fails closed to 1 otherwise (no engine-side reduction, per the design doc's explicit rejection of a min(host,engine) rule). capability-validator.cjs gains parity validation for the new field (optional, positive safe integer or the "undocumented" sentinel — mirroring isolation's "added after existing descriptors" treatment). gsd-tools.cjs gains a new `query dispatch-capacity` route, a pure-read sibling of `dispatch-isolation` with no side effects: live env (GSD_DISPATCH_MAX_CONCURRENCY) > descriptor > fallback-to-1 precedence. All 19 capabilities/*/capability.json descriptors now declare dispatch.maxConcurrency: claude carries the one cited value (20, per code.claude.com/docs/en/sub-agents); the other 18 carry "undocumented" (not yet researched for this axis). * docs(#3673): document dispatch.maxConcurrency and add its citation row to the capability matrix Updates docs/reference/host-integration-interface.md's dispatch struct entry (also backfilling the previously-undocumented isolation/backgroundDispatch sub-fields found stale in the same table) and adds fail-closed/live-transport precedence prose for the new maxConcurrency field. Adds a dispatch.maxConcurrency row (with citation) to all 19 host sections in docs/reference/host-integration-capability-matrix.md — required for tests/host-integration-descriptors.test.cjs's kimi-code matrix-parity check, which asserts every declared dispatch sub-axis is documented there. Updates docs/how-to/add-or-update-a-host-integration.md's dispatch checklist and example descriptor block to mention maxConcurrency (and, likewise backfilling a stale gap, isolation/backgroundDispatch). * fix(#3673): extract shared maxConcurrency validator, drop dead reserved-name check Exports isPositiveSafeInteger from src/host-integration.cts as the single source of truth for the dispatch.maxConcurrency positive-safe-integer contract; negotiateHostCapabilities and gsd-tools.cjs's routeDispatchCapacity now both call it instead of independently reimplementing the same predicate. Also removes the __proto__/constructor/prototype reserved-name branch from capability-validator.cjs's maxConcurrency check — copy-pasted from the string-enum fields above it, but unreachable for a numeric field (the generic positive-safe-integer branch already rejects any string) and absent from maxDepth, the field the code's own comment claims to mirror. --------- Co-authored-by: sim <sim@local>	2026-09-01 21:39:12 -04:00
Tom Boucher	62b0d939b6	feat(#3274): make reviewer-lane timeout configurable via timeoutConfigKey (#4083) ... * feat(#3274): make reviewer-lane timeout configurable via timeoutConfigKey Add an optional `timeoutConfigKey` field to the reviewer lane descriptor, resolved in `resolveLanePlan` at invocation time and falling back to the frozen `timeoutFloorMs` when unset or invalid, in the same spirit as the existing `promptBudgetKey`/`modelConfigKey` fields. All 12 shipped lanes declare `review.timeouts.<slug>` on both surfaces (the descriptor and their capability.json manifest), validated by capability-validator.cjs. For the antigravity lane, the native `agy --print-timeout` flag — previously a second hardcoded literal (`540s`) independent of the outer cap — is now derived from the same resolved outer timeout in `antigravityArgv`, preserving the existing 60-second buffer relationship (ADR-2782 D6: the outer bound is declared data, the inner one is handler-owned). The antigravity default timeoutFloorMs stays at 600s per the maintainer's disposition; users raise it through the new config key instead. * docs(#3274): document review.timeouts.* and extract resolveTimeoutMs helper Address code-review findings on the timeoutConfigKey change: extract the inline timeout-resolution logic into a named, exported, directly-tested resolveTimeoutMs helper (matching the file's existing configString/ normalizeHost convention); document the new review.timeouts.* federated config keys in docs/CONFIGURATION.md, docs/reference/capability-manifest.md, and docs/how-to/ship-a-reviewer-lane.md; add the changeset fragment. * fix(#3274): resolve native antigravity timeout in resolveLanePlan, not the runner gsd-test caught two design mistakes in the prior commits: 1. SpawnPlan.argv is documented and tested as fully resolved by resolveLanePlan (model/effort/output/prompt already folded in) — leaving the antigravity '{{nativeTimeout}}' marker unresolved until the runner's antigravityArgv violated that contract and broke tests that read plan.argv directly (tests/antigravity-reviewer.test.cjs, tests/review-default-reviewers-workflow.test.cjs). Fix: '{{nativeTimeout}}' is now a fifth ARGV_PLACEHOLDER member, resolved by resolveLanePlan itself via the new nativeTimeoutToken() helper, exactly like the other four. antigravityArgv reverts to its pre-#3274 four-argument form. Also missed updating capabilities/antigravity/capability.json's invoke.args to match the descriptor, which broke the manifest/descriptor parity test. 2. tests/reviewer-config-federation.test.cjs enforces a deliberate, narrow invariant (#3691 narrows #2797): qwen, cursor, and coderabbit — the three lanes with neither a model flag nor a host — may own no config key beyond their own prompt-budget key. Adding review.timeouts.<slug> to all 12 lanes violated it. Fix: those three keep timeoutConfigKey: null and own no review.timeouts.* key, matching their existing modelConfigKey: null. The other 9 lanes are unaffected. * chore(#3274): backfill changeset PR number (pr:0 -> 4083) --------- Co-authored-by: sim <sim@local>	2026-08-30 14:26:02 -04:00
github-actions[bot]	4a705cf19f	chore: sync next package version to 1.12.0	2026-08-30 03:07:03 +00:00
Tom Boucher	03b7125293	enhance(#3909): a probe that could not run no longer asserts a verdict (#3944) ... * test(#3909): failing-first suite for the fabricated probe fallbacks Binds the four fabrication sites found by executing the surfaces (ADR-3889 failure class (c)), each with a positive control so an over-firing fix goes red: - the blocking api-coverage.verify-pre gate certifying "no external-API integration" from a zero-byte phase scope - the assumption-delta query route scanning an unresolvable phase section as the empty string and reporting it as an examined negative - both capability fragments' probe fallbacks, which append a fabricated verdict rather than replacing, and fire on the legitimate exit-1 negative Verification runs on the remote runner. Refs #3909 * enhance(#3909): a probe that could not run no longer asserts a verdict ADR-3889 Phase 5. Four sites turned a failed or unexamined probe into a confident negative; each now reports what it could not establish. - check api-coverage.verify-pre: a phase with no plan body and no roadmap section ran detection over zero bytes and PASSED the blocking seal gate, certifying "no external-API integration" from input it never read. It now holds with scope_unavailable. The discriminator is bytes examined, never signals found, so a phase whose plans are real and simply carry no API vocabulary passes exactly as before. - query assumption-delta scan: an unresolvable phase section was scanned as the empty string and reported as an examined negative. It now returns {skipped, reason: phase_unresolved}, still at exit 0 — an ADR-2980 degraded result in the payload, leaving the gsd-tools exit projection to P8. - both capability fragments: `|| echo '{"detected":false}'` appended rather than replaced, and fired on the legitimate exit-1 negative, so a correct answer and an honest skip both arrived as two concatenated objects. They now keep the probe's own payload and manufacture only an explicit probe_unavailable skip when the probe produced nothing at all. Every registered outcome is more restrictive on a blocking gate, so this can turn a false green red and never a red green. Docs: FEATURES 156, CONFIGURATION (both keys), references/api-coverage.md seal-time outcome table, and a new how-to for the reason-code vocabulary. Verification runs on the remote runner. Closes #3909 * test(#3909): correct the stale unknown-phase assertion `unknown phase → detected:false, no throw (graceful)` scanned phase 999 against a two-phase roadmap and asserted `detected === false`. That pinned the fabrication as intended behavior: the phase does not exist, so the detector was handed the empty string and its "no core assumption changed" answer described nothing that was ever read. It now asserts the skipped-with-reason shape. The graceful-degradation contract the test was actually protecting — the query succeeds and does not throw on an unknown phase — is unchanged. Found by code review, not by the author. Refs #3909 * docs(#3909): author the FEATURES entry in its generator source `docs/FEATURES.md` is generated by `scripts/gen-features.cjs` from the per-feature fragments in `docs/features/`. The API-coverage entry was edited in the generated file, so the next regeneration silently dropped it. The text now lives in `docs/features/api-coverage-gate.md` and `docs/FEATURES.md` is regenerated from it, leaving the shipped file byte-identical and its content actually derivable. Caught by `lint:generated-sync`. Refs #3909 * test(#3909): bind the skip to "not found", and pin the discriminator The first verification run went red on one case, and the case was wrong rather than the code. `getRoadmapPhaseWithFallback` returns `null` for an unknown phase and for a missing ROADMAP.md, but for a section whose body is whitespace-only it returns the heading line alone — which is not empty. So a body-less section WAS found, and reporting `detected:false` over its heading is a real negative, not a fabrication. The test had assumed the resolver yielded `''` there. Correcting the test rather than the resolver keeps `skipped` bound to the distinction the issue asks for — found versus not found — and avoids diverging `assumption-delta scan` from `roadmap.get-phase`, which the fragment documents as sharing one resolver. Also adds the seeded property the test matrix had promised: for any plan body, the scope read back is whitespace-only exactly when the body was. That pins the gate's discriminator to bytes examined, so it cannot quietly become "no signals found", across unicode whitespace and CRLF. `docs/INVENTORY.md` picks up the reference doc's new seal-time outcome table — surfaced by the co-change gate, not by a lint failure. Refs #3909 * chore(#3909): backfill the changeset PR number Refs #3909 --------- Co-authored-by: sim <sim@local>	2026-08-27 15:50:12 -04:00
Tom Boucher	c5f2b94b27	enhance(#3907): gates report no-input instead of a verdict they never reached (#3932) ... * feat(#3907): gates report no-input instead of asserting a verdict they never reached The three stdin-reading gates bound 2 to a stdin read error only, with no arm for stdin closed at zero bytes - so empty input flowed into the detector, found nothing, and exited 1, which each module's own comment defines as a negative verdict. An unset PHASE_SECTION made the UI gate assert the phase has no UI. Empty and whitespace-only input now exit NO_INPUT, and a read error exits UNAVAILABLE rather than a locally-invented 2, both resolved through the registry and delivered by terminateNow. The exit code was only half of it: under --json the same input emitted {detected:false}, byte-identical to the fabricated payload #3909 exists to fix, and the blocking coverage gate reads that payload. Empty input now emits the in-tree {skipped:true,reason} form with no detected key at all. teams-status is excluded: it never reads stdin and has no invented 2, so the four-module framing in the issue and ADR is wrong. The dead root bin/lib/ui-safety-gate.cjs is deleted - no installer reference, no workflow invocation, and the live fallback chains are for other modules. Its removal restores the unit tests to the module that actually ships; they had been asserting the stale copy's two-field shape, which is why it drifted unnoticed. * fix(#3907): drive gate tests through the process seam, and make removed-but-needed basename-precise CONTRIBUTING requires every subprocess go through tests/helpers/process-seam.cjs; two of the three gate suites hand-rolled spawnSync while the third, added in the same change, used runNode correctly for the identical injection case. Converted the blocks this change added, leaving pre-existing ones alone. Deleting one of two files sharing a basename made lint-removed-but-needed report 14 references that were all to the surviving canonical module - the false-positive class its own docstring names. It now matches on the deleted file's full path when a surviving file shares its basename, which is more precise rather than weaker: a genuine full-path reference still fails, and behaviour is unchanged when no basename collides. It immediately caught a docstring on this branch that spelled the deleted path. * test(#3907): update the one existing assertion that pinned the old empty-stdin verdict A pre-existing test asserted exit 1 on empty stdin - the defect this phase removes - and was missed because the change added new blocks without auditing existing ones pinning the old contract. Audited the rest: the other three status-1 assertions in that file all feed real input and are the genuine-negative controls that must keep returning 1, so exactly one was stale. The retired 2 is gone from the describe's contract comment too. * chore(#3907): backfill changeset pr number --------- Co-authored-by: sim <sim@local>	2026-08-27 11:37:31 -04:00
Tom Boucher	39673ae9ff	fix(#3738): antigravity global skills/agents install to ~/.gemini/config (#3921) ... * test(#3738): antigravity global skills/agents must resolve under ~/.gemini/config Regression tests (RED first): --skills-root and gsd-tools query surfaces, install-plan dest dirs, and converter skills-path rewrite. * fix(#3738): antigravity global skills/agents install to ~/.gemini/config Antigravity's machine-local discovery scans ~/.gemini/config/{skills,agents}; the configHome (~/.gemini/antigravity) is deprecated for artifacts. Declare the ADR-1239 skills/agents 'home' override on the antigravity global layout — the same mechanism codex uses (.agents) — and divert ~/.claude/skills/ references in converted global content to ~/.gemini/config/skills/. configHome, settings, probe/migration semantics, and the local .agents layout are unchanged. * fix(#3738): retire deprecated configHome artifacts via installer migration 010 Next install converges an existing antigravity install: manifest-managed skills/gsd-*/ and agents/gsd-*.md under the configHome (a location AGY does not scan) are removed — modified files backed up first, unmanifested and non-gsd entries preserved — and now-empty containers retired. Global scope only; the local .agents surface is live. Docs + inventory updated. * fix(#3738): converter sync in bin/install.js, harness emit-root coverage, migration baseline - bin/install.js converter gains the same ~/.claude/skills → ~/.gemini/config/ rewrite as src (ADR-1508 dual copy must stay in sync). - Parity-manifest walk covers home-override emit roots (extraEmitRootsFor) so antigravity's emitted skills/agents stay differential-visible at their new install root; install-tree fixture regen confirms an unchanged key set. - skills-from-commands rule declares the antigravity converter as a runtime-scoped transform; one ack fragment covers the identity-classed workflow whose antigravity copy embeds the old skills path. - Migration 010 checksum baseline + home-override set doc updated; existing tests updated to the #3738 contract (global dest, golden parity via layout dest, integration expectations). * fix(#3738): tolerate an absent extra emit root on baseline-side measurement The base tree's installer predates the home override, so <HOME>/.gemini/config does not exist there; walk() threw ENOENT and the in-job baseline build failed. An absent extra root is the legitimate pre-override shape — skip it. * fix(#3738): review findings — manifest agents root, bare skills-path rewrite, guard comment - writeManifest resolves the agents-kind home override (_kindDestDirSafe), so the manifest records agents at their actual install root and drift detection keeps working (isolated review finding 1, major). - Converter bare forms ~/.claude/skills and $HOME/.claude/skills (no trailing slash) divert to ~/.gemini/config/skills instead of falling through to the retired configHome path (finding 2). - real-home-guard comment updated: antigravity's global agents kind is the first agents-kind home override (finding 3, doc-only). - Regression tests for both behavioral findings. * chore(#3738): changeset fragment (pr number backfilled after PR creation) * chore(#3738): backfill changeset PR number (3921) * fix(#3738): sandbox HOME in tests that install antigravity global artifacts antigravity is the first home-override runtime in the golden-parity and skills-wrapper suites (codex is not in their runtime lists), so those tests never needed HOME sandboxing — the real-home guard now (correctly) refuses their un-sandboxed global installs on CI, where HOME is the passwd home. * fix(#3738): stop the K3 sequential-sandbox env leak; sandbox L2's home-override plans K3's two back-to-back sandboxHome calls leave HOME pointing at the first sandbox once the after-hooks restore (each call saves the env as it found it, so the second saves the first's sandbox as 'original'). On the windows matrix that leaked gsd-k3-qwen-* home into the L2 property, whose antigravity/global run then (correctly) refused via the #3712 real-home guard — antigravity is the runtime that made L2's plan escape into os.homedir(). K3 now manages the env with a single restore; L2 sandboxes HOME per run, mirroring L1. * fix(#3738): L2 property's HOME sandbox must exist on disk The #3712 guard's sandbox exemption fails closed when identify(effectiveHome) is 'absent' — L2 never created its configDir, so on the windows matrix (tmpdir under the real home) the antigravity/global run refused even with HOME sandboxed. Create the per-run sandbox dir and clean it up. --------- Co-authored-by: sim <sim@local>	2026-08-27 02:24:03 -04:00
Tom Boucher	a3d5841117	fix(#3714): deliver an explicitly pinned model to the Codex worktree executor, and drop an unusable one (#3891) ... * test(#3714): failing-first coverage for the dropped Codex worktree model override Pins the argv contract for the orchestrator-worktree process dispatch: an explicit model_overrides pin must reach the child as --model, while an unpinned, empty, inherit, or profile-only configuration must emit no flag at all. Five of the eight matrix rows are CONTROLS that pass before the fix. They carry the weight here because this is an over-emission bug waiting to happen: resolve-model returns 'sonnet' for the unpinned, empty and profile-only cases, so a fix that threads its return value into argv would satisfy the positive row and emit --model sonnet to Codex on every unpinned install -- the documented 400 that ADR-2313 exists to prevent. The controls are what separate the correct fix from the obvious one. * fix(#3714): deliver an explicitly pinned model to the Codex worktree executor resolveOrchestratorExec had no model input at all -- the descriptor carried only command/args/cwdFlag/promptFlag -- so a resolved override had no way to reach the spawned process even in principle. The baked gsd-executor.toml could not compensate because this path spawns a process rather than dispatching a named agent. Three parts, and the third is the load-bearing one. The descriptor gains modelFlag (codex: --model), keeping the per-host knowledge as descriptor data exactly as cwdFlag and promptFlag already are, so the scheduler grows no per-host branch. No other runtime declares it. The seam appends [modelFlag, model] and stays MECHANICAL: it does not know the inherit sentinel, does not know which models Codex rejects, and reads no config. Its sibling codex-agent-toml states that rule outright -- callers decide what to strip. Argv order is baseArgs, model, cwd, prompt so the prompt remains the final positional token. Omitting the model is byte-identical to before. A model starting with '-' now fails closed as unsafe_leading_dash_model, the same hazard the prompt and cwd guards already reject and which was silently accepted before. The policy lives at the caller and passes ONLY an explicit, non-sentinel per-agent pin. Passing null as the runtime resolver is what keeps profile and tier derived models out of argv, which is what Codex's session-only model posture requires: the model resolver returns 'sonnet' for the unpinned, empty and profile-only cases, and emitting that revives the documented 400 from #2310/#2311 that ADR-2313 removed. So the gate is the presence of an explicit pin, never that a value came back. * fix(#3714): enforce the real-Codex value policy the sibling surface already applies Review found one root cause behind a BLOCKER, two MAJORs, two MINORs and an argv-injection finding: the dispatch path gated on the PRESENCE of an explicit pin but never applied the VALUE policy that generateCodexAgentToml already applies to the same config key. Textbook generative divergence -- and the parity row I wrote tested resolver parity, not this policy, so it could never have caught it. BLOCKER: a global ~/.gsd/defaults.json model_overrides.gsd-executor of 'sonnet', 'opus' or 'claude-sonnet-4-5' reached codex exec --model verbatim. That is the documented 400 from #2310/#2311, arriving through the explicit-pin door rather than the tier door. The issue asks for an explicit REAL-CODEX pin; real-Codex was unenforced. Now dropped with a warning via the isAnthropicFlavoredModel predicate #3241 single-sourced for exactly this reason. Also: values are trimmed, so a whitespace-only pin is blank rather than --model " "; 'inherit' is matched case- and whitespace-insensitively, so 'Inherit' and ' inherit ' no longer reach the wire; and a value outside a model-id charset is dropped with a warning. That last one closes the injection surface -- .planning/config.json travels with a clone, and values like 'gpt-5 -c approval_policy=never' or a command-substitution value previously reached argv verbatim, where the spawner is an agent writing bash. Every rejection DROPS AND WARNS rather than failing closed. An unusable exec is not degraded, it is fatal: the dispatch step halts the wave after the worktree already exists, so a config typo would have aborted execute-phase. The stale comment claiming it degrades to sequential is corrected. Separately, the seam's own empty-model handling contradicted the committed contract and failed four tests on the remote runner. An absent, null or empty model is not an error -- it means use the host default, the same degradation cwdFlag:null already expresses. Unlike a prompt, where empty is a hang rather than a degraded run, so that one stays fail-closed. Non-string values still fail closed. Tests: the CLI rows were not HOME-hermetic and read the developer's real ~/.gsd/defaults.json, which is precisely the file the BLOCKER is about; HOME and USERPROFILE are now sandboxed per call. Adds the global-pin regression, the injection shapes, the case-variant inherit rows, and a real cross-surface divergence guard. * fix(#3714): close the flag-shaped pin, the case-variant alias, and the warning sink Round two of review found three more defects, two of which both engines reached independently, and all three were mine. The charset allowlist put the dash INSIDE the character class, so a value made only of allowed characters passed the pin policy silently and then tripped the seam's leading-dash guard, producing exec:null. That is the wave-fatal path the whole drop-and-warn design exists to avoid, reachable from a committed config file: -c, --config, -p and --dangerously-skip-permissions all reproduced it. It also regressed hosts with no model flag at all, where a dash pin turned a previously working kimi-code dispatch into exec:null. The first character is now anchored, so a flag-shaped value is dropped and warned like every other rejection, and the comment that claimed this path was unreachable is corrected. isAnthropicFlavoredModel folded case on its substring arm but not on its alias-set arm, so SONNET, Sonnet, OPUS and HAIKU all reached Codex argv while lowercase sonnet was correctly dropped -- the same 400 the drop exists to prevent. The predicate is the one #3241 single-sourced so these surfaces cannot diverge, so folding case there fixes the install-side .toml surface too. The warning wrote the rejected value RAW to stderr. Every value that fails the charset test contains by definition the characters the charset excludes, so it was a guaranteed-reachable raw-to-terminal sink: an OSC sequence in a committed config reached the operator's terminal byte for byte, and truncation could sever an escape before its reset. The value is now sanitized before truncation. Also from review: the allowlist rejected Vertex version pins like text-bison@002, a false positive on a real model id; the policy ran host-neutrally so hosts with no model flag printed a misleading drop warning on every dispatch; the invalid_model branch had no test at all; and the changeset disclosed only that a pin is delivered, not that an unusable one is now dropped with a warning. * fix(#3714): single-source the model-id charset, bound the pin, keep the flag diagnosis Round three found no blocking findings on either engine. These are the three correctness items left in code I added. The charset existed TWICE -- once to accept a pin, once to render a rejected one in the warning -- and the two copies had already drifted inside a single commit: '@' was added to the accept class and not the render class, so a Vertex-shaped value rejected for some other reason rendered as text-bison?002. Both are now derived from one definition, with a parity test asserting every character the matcher accepts survives the sanitizer unchanged, so they cannot drift again. A pin reached argv unbounded. CLAUDE.md documents the hazard: execFileSync aborts on Windows above 32,767 characters of argv. A model id has no reason to be long, so a pin over 200 characters is dropped and warned rather than truncated -- a truncated model id is a different model id. Boundary rows at 199, 200 and 201. The first character is now required to be alphanumeric, so '@evil' and '/c' no longer reach argv. Security rates both inert on codex today, so this is hardening rather than a live defect; it is here because it is one character of regex and resolveOrchestratorExec documents itself as a general descriptor-to-argv seam that other hosts may adopt. Tightening the anchor made the leading-dash branch unreachable and, with it, regressed the diagnosis: '-c' began reporting 'unsafe characters' instead of 'looks like a flag/option'. The dash check now runs before the charset test, which both restores the actionable message for the most likely user typo and keeps the branch live. A test pins the distinction between the three rejection messages so the branch cannot silently die again. * test(#3714): make the charset parity guard actually guard, and remove a false-green trap Review proved by mutation that my parity test could not do what its own comment claimed. It bound the expected character set to a local that was assigned and discarded, and the shared definition was not exported, so widening that definition left the test green. A comment overstating what a test guards is worse than no comment, because the next reader trusts it. The shared body is now exported and asserted equal, and I watched the assertion fail against a widened definition before keeping it. The sanitizer regex was module-scope, carried the g flag and was exported. Production only uses it with replace, which resets lastIndex, so there was no live bug -- but test() on a g-flagged regex alternates between calls, so any future test reaching for it would false-green. The g-flagged copy is now internal to the one call site that needs it and the exported companion carries no flags, which removes the footgun rather than documenting it. Also notes at the shared definition that it is interpolated into both a positive and a negated character class, so only plain characters and ranges are safe to add. No runtime behavior changes: the accept set, the anchor, the length cap, the sanitizer output, the rejection ordering and all three message texts are byte-identical, re-verified through the real CLI. * chore(#3714): backfill changeset pr number * chore(#3714): re-trigger CI after the GitHub Actions outage The workflow runs for this branch were created during the Actions major outage on 2026-08-26 and never got scheduled. They are wedged: GitHub reports them queued, refuses to cancel them, and refuses to rerun them because it believes the workflow is already running. The Tests run is also pinned to a superseded sha, so no test run exists for the current head at all. Actions is operational again and the repo-wide queue has drained, so a fresh push is what creates schedulable runs. This commit is empty on purpose: nothing about the change is being altered, and the verified content is byte-identical to 7860c6ccc. --------- Co-authored-by: sim <sim@local>	2026-08-26 17:01:09 -04:00
Tom Boucher	aaf47c5fc2	fix(#3691): let every reviewer lane take a prompt cap, and make the documented global resolve (#3832) ... * test(#3691): failing-first coverage for the reviewer prompt budget No prompt cap can reach any CLI reviewer lane, by any configuration. Two independent defects compound: all nine `transport: spawn` lanes declare `promptBudgetKey: null`, so `budgetFor` returns on its first line; and the documented global `review.max_prompt_tokens` is advertised in the schema manifest but declared nowhere, so the resolver never materializes it and `budgetFor`'s fallback is dead code. Adds to tests/reviewer-config-federation.test.cjs, which already owns the per-reviewer budget config-set/config-get idiom: - a CLI lane inherits the global cap (RED: reports null) - an http lane with the -1 sentinel inherits the global cap (RED: reports null) - the resolved review surface carries max_prompt_tokens at all (RED: absent) - per-lane overrides the global on a CLI lane - the sentinel boundary: -1 inherits, 0 means do-not-trim and must NOT read as unset, 1 is the smallest real budget — the regression budgetFor's own comment warns about - anti-tightening pins that must stay green: an empty config leaves every lane null, the three existing budgeted lanes are unchanged, and config-set still rejects a per-reviewer key naming something that is not a declared lane - a fast-check property over the resolution contract itself, with -1, 0 and non-finite inputs generated explicitly rather than left to chance Every row was reproduced by hand against the real CLI before being written, so the RED/GREEN split is observed rather than predicted. Refs #3691 * fix(#3691): let every reviewer lane take a prompt cap, and make the global resolve No prompt cap could reach any CLI reviewer lane, by any configuration. Two independent defects compounded. The nine spawn-transport lanes — claude, coderabbit, antigravity, cursor, gemini, codex, kimi-code, opencode, qwen — declared `promptBudgetKey: null`, so `budgetFor` returned on its first line and `review-lane plan` reported `promptBudget: null` no matter what was configured. Each now declares `review.max_prompt_tokens_per_reviewer.<slug>` with the same `-1`-is-unset sentinel the three local-server lanes already use. Separately, the central `review.max_prompt_tokens` was listed in the schema manifest's validKeys and documented as a supported setting, but declared nowhere — the resolved surface is built from capability declarations plus the defaults manifest, and neither carried it. `configGet` returned undefined and `budgetFor`'s documented fallback was dead code. It is now declared with a `null` default, exactly as docs/CONFIGURATION.md already specified, so the default behavior is unchanged: nothing configured means nothing trims. Two things the diagnosis had not predicted, found and fixed while implementing: - `REVIEWER_LANES` in src/review-lane-descriptor.cts is a second, hardcoded registration site that `mergeReviewerLanes` prefers over the capability registry on a slug collision. Editing only the capability files left every CLI lane still null. Both sites now agree. - The generated `gsd-core/bin/lib/capability-registry.cjs` was stale and masked the capability edits; regenerated with `npm run gen:capability-registry` rather than hand-edited. docs/CONFIGURATION.md said "Only lanes that declare a budget key accept one — today ollama, lm_studio and llama_cpp". That is false as of this change and is corrected rather than left to rot. The trim-versus-refuse question the issue raises is deliberately not taken up here: the refusal path already exists for the case that matters — a reviewer whose minimum set exceeds its budget is skipped rather than sent a misleading prompt — and trimming above that floor is the documented, shipped design of the feature. Changing it would alter behavior for the three lanes that already work, which is not what the issue asks for. Fixes #3691 * fix(#3691): document the new global and narrow an invariant this change obsoleted The full suite surfaced two consequences of giving every CLI lane a budget key. `review.max_prompt_tokens` entered CONFIG_DEFAULTS without a matching entry in the planning-config reference, which config-field-docs guards. Documented, including the sentinel semantics a reader needs: a per-lane value overrides the global, `-1` means unset and inherits it, and `0` means "do not trim that lane" and is not unset. The #2797 federation guard asserted that "a lane with no model flag and no host owns no config keys". That held only because budget keys existed solely on the three local-server lanes, all of which have hosts. A lane can now legitimately own a config key for a third reason, so qwen tripped it. The assertion is narrowed rather than weakened: such a lane must still own no model key and no host key, and may own at most its own `review.max_prompt_tokens_per_reviewer.<slug>` — never another lane's. That is strictly more specific in the dimensions that still matter. Proven to still bite: hypothetically giving qwen a `review.models.qwen` key fails it with `model/host: review.models.qwen`. The name and comment cite #3691 for why the premise changed, so a reader sees a deliberate narrowing, not erosion. Checked the sibling assertions in that describe block; the other three do not rest on the obsolete premise and are untouched. Refs #3691 * fix(#3685): port the write-flag content-change contract to its three sibling sites #3685 fixed `phase complete`'s `roadmap_updated` / `state_updated`, which reported `fs.existsSync(path)` rather than whether the transaction wrote anything. Three sibling sites carried the identical defect and are ported here. - `cmdPhaseRemove` reported `roadmap_updated: true`, hardcoded. `updateRoadmapAfterPhaseRemoval` now returns whether the content changed and the flag reports it. #2640/#2974 already fixed `state_updated` at this same call site and left this one behind, so the correct shape was adjacent. - `cmdMilestoneComplete` reported `state_updated: fs.existsSync(statePath)` — byte-identical to #3685's bug in a different command. - `cmdMilestoneComplete` reported `milestones_updated: true`, hardcoded, never consulting the MILESTONES.md write. `gsd-core/workflows/remove-phase.md:100` extracts `roadmap_updated` for display and never branches on it, so the flip from always-true to content-based changes no workflow behavior. Verified by reading the step, not assumed. One trap found while implementing: the obvious in-memory `finalContent !== originalStateContent` comparison — copying `cmdPhaseComplete`'s shipped shape verbatim — gives a FALSE POSITIVE for milestone completion. `platformWriteSync` normalizes Markdown at write time, and the milestone-closure transform regenerates `## Current Position` fresh on every call, so its pre-normalize output always differs from the already-normalized file on disk even when the persisted bytes are identical. The comparison is therefore made against the post-write on-disk content. `cmdPhaseComplete`'s own comparisons are left untouched — their repeat-no-op tests pass, so they are not exposed to this artifact. `milestones_updated` has no reachable no-op: the MILESTONES.md write unconditionally appends an entry every call. Only the true direction is pinned, documented inline rather than faked with a passing test. Refs #3685 * fix(#3685): compare write-flag content through the writer's own normalizer An independent reviewer disproved a claim made while porting #3685's contract to its sibling sites: that `cmdPhaseComplete`'s comparisons were not exposed to the Markdown-normalization artifact already diagnosed in `cmdMilestoneComplete`. `platformWriteSync` normalizes on write — CRLF stripped, blank-line runs collapsed, a blank line inserted after a heading, a single trailing newline enforced. Every flag that compares the PRE-normalization in-memory string against the on-disk pre-image can therefore report a change when the persisted bytes are identical. `cmdMilestoneComplete` had been worked around by re-reading the file after the write; the other sites compared raw strings. All of them now go through one exported seam, `contentChangedAfterNormalize(filePath, before, after)`, which normalizes both sides exactly as the writer does. That removes the extra disk read the milestone workaround needed, and makes the sites agree by construction rather than by four independent implementations of one rule — the divergence the repo names as an anti-pattern. Reachability, stated precisely rather than uniformly: the seam is load-bearing at `cmdPhaseComplete`'s `roadmapUpdated`, `requirementsUpdated` and `stateUpdated`, where section-rewrite logic genuinely regenerates content into a different-but-normalization-equivalent shape. At `updateRoadmapAfterPhaseRemoval` it is defense-in-depth: the no-match branch never reassigns `content`, so the raw comparison was already correct there. The first analysis claimed the reverse; this is the corrected finding. Also fixes an unsound test premise the remote suite caught. The byte-identity precondition in `roadmap_updated is false when ROADMAP.md comes out byte-identical` asserted against a hand-authored, un-normalized fixture — so the very first write reformatted it and the file could not come back identical. The fixture is now written already-normalized, so the assertion compares a normalized pre-image against a normalized post-image and still fails if the flag regresses to a hardcoded `true`. Not platform-specific; it reproduces on macOS too, and the earlier local check simply never exercised it. The sibling true-direction and milestone tests were checked for the same premise and do not share it — they assert `notEqual`, or compare two post-write states produced through the same normalizing seam. Refs #3685 * chore(changeset): backfill PR number for #3691 fragment --------- Co-authored-by: sim <sim@local>	2026-08-24 19:39:51 -04:00
Tom Boucher	2f86278b5e	fix(#3003): opt-in mechanism for intentional deletions in worktree.cleanup-wave (#3757) ... * test(#3003): failing-first suite for declared deletions in cleanup-wave Binds the guard's opt-in before it exists, so the suite is RED against next. The rows that carry the weight are the over-authorization set: a directory declaration must not authorize its children, a glob declaration must authorize nothing, and a declaration must not act as a string prefix of another path. Each of those BLOCKS, and each would PASS under a prefix, glob, or startsWith matcher — which is how a path list quietly degrades into the boolean opt-in #3003 explicitly rejected. The glob row matters most: declaredScopePrefix already returns null ("matches everything") for a glob-leading pattern, correct for the advisory it serves and catastrophic for a gate. Also pinned: a failed deletion check blocks on its own reason rather than being filtered into a pass; the block detail names only the undeclared residue so the operator is not misdirected by paths that were fine; an entry with no declaration blocks exactly as before; junk and non-array declarations do not authorize; and a blocked entry still isolates rather than aborting the wave (#2852, which must stay fixed). Two advisory rows cover an interaction found while designing: git diff --name-only includes deleted paths, so without unioning the declaration into the #2596 scope check, authorizing a deletion would raise SCOPE_OUT_OF_DECLARED against the very path just authorized. A seeded property states the whole invariant the three over-authorization rows sample: a deletion merges iff its normalized path is in the declared set. * feat(#3003): declared deletions opt-in for the cleanup-wave guard The deletions guard blocked the merge-back of any executor branch whose diff removed a file, with no way to say a removal was intended. A plan that folded one test file into a sibling could not be merged by the tool meant to merge it, forcing a manual --no-ff outside the tool -- strictly less safe than what the guard protects against. A plan now declares removals in its own frontmatter (files_deleted), and that list rides the same path files_modified already travels: plan-document parse -> phase plan JSON -> the per-plan worktree gate -> record-agent/create --deletions -> declared_deletions on the manifest entry -> the guard. The guard blocks only the deletions NOT in that list. A path list rather than a boolean, per the pinned decision: a boolean disarms the guard for the whole entry, so an unexpected deletion riding along with a declared one would pass unnoticed. Matching is exact after the module's shared normalizer -- never a prefix, never a glob. Both would let one declaration authorize a whole set, which is the mass-deletion accident the guard exists to catch. That also means declaredScopePrefix is deliberately NOT reused here: it returns null ("matches everything") for a glob-leading pattern, which is right for the advisory it serves and would silently disarm a gate. The block detail now carries only the undeclared residue, so an operator is not sent looking at paths that were fine. A failed deletion check still blocks on its own reason and is never filtered into a pass. A blocked entry still isolates rather than aborting the wave (#2852). The #2596 scope advisory unions the declaration into its declared set -- git diff --name-only includes deleted paths, so without that, authorizing a deletion would immediately warn that the same path was out of declared scope. Optional and additive throughout: files_deleted is absent from PLAN_REQUIRED_FIELDS, a manifest entry without declared_deletions keeps the original unconditional block, and omitting --deletions leaves the on-disk entry shape untouched. Supersedes the spent #2856 emitted-drift ack entry for execute-phase.md, the same supersede that entry performed on #3370 and #3370 on #3324. * fix(#3003): wire --deletions on every dispatch surface, not just one Review found the feature inert on two of three dispatch paths. execute-phase.md (harness inline) passed --deletions, but the orchestrator-worktree path (executor-isolation-dispatch.md, worktree.create) and the Fleet-parallel batch path (capabilities/claude-orchestration/fragments/execute-wave-pre.md, worktree.record-agent) still passed only --files. A plan declaring files_deleted would have merged on one path and been blocked on the other two -- the exact bug #3003 exists to fix, left unfixed where most of the isolation actually runs. Worse, per-plan-worktree-gate.md already claimed --deletions was passed 'on the same worktree.record-agent / worktree.create calls', which was false for both untouched sites. A doc asserting coverage that does not exist is how a gap survives review. All four surfaces now pass the flag, verified by sweeping every .md under gsd-core/, capabilities/, commands/, skills/ and agents/ that invokes worktree.record-agent or worktree.create: each one that passes --files now also passes --deletions. The isolation-dispatch note explains why this flag, unlike --files, is not advisory -- omitting it does not skip a check, it blocks a merge the plan declared. Regenerates capability-registry.cjs, which the fragment edit made stale. Neither newly-grown file needs an emitted-drift ack: executor-isolation-dispatch.md sits under workflows/execute-phase/steps/ and execute-wave-pre.md under capabilities/, both outside currentSizes()'s non-recursive scan of gsd-core/workflows/ and agents/. * docs(#3003): document files_deleted where a plan author will actually find it The feature's entire user surface is one plan-frontmatter field, and the canonical reference for that frontmatter -- docs/reference/plan-md.md, the table that documents every other key -- never mentioned it. A field nobody can discover ships as a field nobody uses. Adds the files_deleted row and an example entry in all five locales (en, ja-JP, zh-CN, ko-KR, pt-BR), stating the property that makes the opt-in safe: matching is exact per path after separator normalization, with no globs and no directory prefixes, so a declaration can never authorize more than it literally lists, and omitting the field keeps the guard's original unconditional block. Also corrects two claims in the scope-conformance how-to that this change made false. Its opening paragraph described the recorded declared scope as files_modified alone; declared_deletions is now unioned into that comparison. Its "Renames are not detected specially" bullet asserted the deletions guard blocks any entry whose diff contains a deletion, full stop -- which was the whole point of #3003 and is no longer true. Reworked to say what now decides a rename's fate: declare the old path in files_deleted and both halves become ordinary paths for the advisory check, which is also why the old path needs no separate files_modified entry. Documentation that describes the pre-change behavior of the thing being changed is worse than no documentation, because a reader trusts it. * fix(#3003): close every review finding on the declared-deletions opt-in Two independent isolated reviewers, correctness and security. Neither found a blocker; both found real defects, and the directive treats a finding at any severity as blocking. All of them are fixed here. MAJOR -- the submodule worktree gate could not see a deletion-only plan. per-plan-worktree-gate.md intersected $SUBMODULE_PATHS against $PLAN_FILES alone, while $PLAN_DELETIONS was extracted and then never used. Before files_deleted existed, a path had to appear in files_modified to be planned at all, so the gate saw it; the new field plus the new docs telling authors a deleted path needs no files_modified entry opened a hole where a plan whose only submodule touch is a removal kept worktree isolation on -- the exact case #2772 disabled it for. Both channels now feed the intersection. Note the posture is deliberately the OPPOSITE of the cleanup-wave guard: there the channels stay apart because a deletion AUTHORIZATION must never be inferred; here they merge because a safety fallback must never MISS a touch. MAJOR -- same-wave conflict detection could not see a deletion. The planner's implicit-dependency rule compared files_modified only, so plan A editing src/x.ts and plan B declaring files_deleted: [src/x.ts] scored as conflict-free and ran in parallel: one branch removing what the other is writing, which is the sharpest conflict there is. Overlap is now computed across both channels. MINOR (both reviewers, one root cause) -- the advisory union gave one field two matching rules. declared_deletions was unioned into the scope list handed to planWaveScopeConformance, which reads it with prefix-and-glob semantics. So a field that is exact-match-only at the gate silently became wider at the advisory: ["*.md"], inert at the gate, yielded a null prefix meaning "matches everything" and muted the advisory completely, and ["src"] muted all of src/. The union also activated the advisory on plans that declared no modification scope at all, warning on every modified path. Replaced with subtraction from the findings, gated on files_modified alone. One field, one rule, everywhere. MINOR -- core.quotepath made the feature silently inert for non-ASCII paths. git emits "tests/\303\251.ts" C-escaped and quoted, which never equals the declared plain path, so a correctly declared deletion of tests/é.ts would block forever with nothing pointing at the encoding. Both diffs now pass -c core.quotepath=false. NIT -- flag() consumed a following flag as a value, so --deletions --files x swallowed --files and dropped both. Now treated as a missing declaration, which fails closed. Fixed at both call sites; the helper is duplicated verbatim in cmdWorktreeRecordAgent and cmdWorktreeCreate and leaving one would reintroduce it. TEST -- one test passed for the wrong reason. "a declared deletion is in scope for the advisory" asserted only that warnings omit the deleted path; under a full revert the entry blocks first, warnings come back empty, and the negative assertion passes anyway. It now asserts the entry actually merged, which is the load-bearing half. Four regressions added, one per fix above. Docs corrected rather than extended. The rename bullet in the scope-conformance how-to claimed a rename whose delete side is undeclared never reaches the advisory. Verified false: git's rename detection is on by default, so a pure rename is a single R entry that appears in no --diff-filter=D output and was never gated, before or after #3003. Only a rename that edits enough to fall below the similarity threshold decomposes into add+delete. The pre-existing sentence made the same wrong claim; this restates it correctly instead of sharpening the error. The localized plan-md.md reference edits are reverted: the PR template requires docs content added here to be English, and the translations already lag by three fields, so English-only is the repo's standing posture, not an oversight. Agent-file size caps respected: gsd-planner.md is XL-tier by bytes but carries a separate 49152-LF-CHAR cap asserted by four suites, so its edit is deliberately terse and lands at 49141 with 11 chars of headroom, with the rationale moved to docs/reference/plan-md.md, which has no cap. gsd-plan-checker.md lands at 49107 bytes, 45 under the LARGE cap. Both acks merged into the existing fragments that already name those paths, since two ack sources may never name the same path. * fix(#3003): decode git's path quoting instead of changing the git argv The previous commit's non-ASCII fix turned the remote suite red: 44 failures, 42 of them "unexpected git call: -c core.quotepath=false diff --diff-filter=D --name-only ...". The suite's git mocks match on exact argv, so adding two flags to the deletions diff and the advisory diff invalidated every existing fixture in tests/worktree-safety.test.cjs. Rewriting dozens of fixtures to accommodate one flag would be paying a large Hyrum's-law bill to fix a small defect. Both execGit calls are reverted to their original argv. The C-quoting is now decoded in normalizeScopePath instead, via a new decodeGitQuotedPath helper. That is the better fix on its own merits, not merely the cheaper one: the git argv is untouched so no fixture moves, the decode lands on the ONE normalizer already applied to both sides of the comparison so the declared and reported paths cannot disagree, and it holds regardless of the user's own core.quotepath setting rather than only when we remember to override it. A value not wrapped in a leading AND trailing quote is returned completely untouched, so the plain-ASCII path -- the overwhelmingly common case -- is byte-identical to before. Escapes decode to BYTES collected into a Buffer and UTF-8 decoded only at the end, because \303\251 is two bytes forming one character and decoding them separately yields mojibake. Malformed input never throws: a trailing lone backslash or a short octal escape degrades to the literal character, since one bad path must not take down a cleanup wave. Caught while reviewing the helper: the non-escape branch pushed a UTF-16 code unit rather than UTF-8 bytes. Git always escapes non-ASCII so its own output was fine, but this normalizer runs on the DECLARED side too, and an author may write a quoted path holding a literal é -- pushing 0xE9 alone is invalid UTF-8, so the declaration would decode to a replacement character and silently stop matching. That is precisely the failure this change removes, reintroduced on the other side of the comparison. Now converts whole code points, surrogate pairs intact. The other 2 failures: tests/parallel-dependent-plans.test.cjs pins the exact unbackticked substring "files_modified overlap" in gsd-planner.md, and rewording that comment to "declared-scope overlap" deleted it. The comment is restored verbatim and the files_deleted change rides in the pseudocode and the Rule sentence instead. Recorded in the ack fragment so the next contributor does not rediscover it the same way. Four regression tests cover the decode through the public cleanup-wave seam (the helper is module-private): a declared non-ASCII deletion merges against a C-quoted git report, the symmetric case where the DECLARATION is the quoted form, an undeclared non-ASCII deletion still blocks with the residue naming the decoded path an operator can act on, and a path merely containing a quote is left alone. Plain ASCII was already covered and is not duplicated. * fix(#3003): revert the leading-dash flag guard, the review nit was wrong The remote suite came back with 2 failures, down from 44, and both point at the same thing: tests/worktree-safety.test.cjs:7045 already pins the opposite contract, deliberately. test('a flag-shaped --files value is not re-parsed as a flag', ...) recordAgent(['--files', '--branch']) -> files_modified === ['--branch'] -> branch === 'worktree-agent-a1' ("the real --branch value must be untouched") So consuming the next argv element positionally, whatever its shape, is the tested intent of this parser, not an oversight. The security reviewer's nit claimed --deletions --files x would "swallow --files and drop both". It does not: each flag runs its own indexOf, so --deletions records the literal '--files' while --files independently still resolves to x. And that literal is a path git never reports as deleted, so it authorizes nothing -- already fail-closed with no guard at all. The guard bought no safety and silently changed --files behavior along the way, outside this issue's scope. Reverted at both call sites, which are byte-identical again, along with the test asserting the reverted behavior and the docs sentence describing it. The nit is recorded as REJECTED in the review artifact with the reasoning above, rather than as fixed -- a finding that turns out to be wrong should leave a trace of why, or the next reviewer files it again. docs/CLI-TOOLS.md now states the positional-read behavior plainly instead, so the next person meets it as documented intent rather than rediscovering it through a red suite. * chore(#3003): backfill changeset pr number to 3757 * test(#3003): cover parsePlanDocument's filesDeleted branch to clear the mutation gate CI's Stryker shard for plan-document failed at 73.28 against a break threshold of 75: 170 killed, 62 survived, 232 total. Eight of those survivors are the filesDeleted block this issue added to parsePlanDocument, which shipped with no direct coverage at all -- the field was exercised end to end through the cleanup-wave tests, but the parser itself was never called with a plan that declares it, so every mutant in the block lived. Four tests, each pinned to specific mutants rather than written for coverage percentage: - absent key yields exactly [] -- kills the array-literal seed (["Stryker was here"]) and the `fmDeleted = true` conditional, which would otherwise produce ["true"] - a scalar underscore `files_deleted:` wraps into a one-element array -- kills `fmDeleted = false`, the `&&` logical-operator swap, the `fm[""]` string mutation on the first operand, the emptied if-block, and the ternary's non-array branch - an array-valued hyphenated `files-deleted:` maps element-wise -- kills the `fm[""]` mutation on the SECOND operand (only reachable when the legacy hyphen alias is the one carrying the value) and the ternary's array branch - an empty list yields [] -- boundary case, and a genuinely distinct one from the absent key: [] is truthy in JS so it ENTERS the if, and only Array.isArray's true branch mapping over nothing produces the same [] Threshold arithmetic: 174 of 232 are needed for 75%, and these take it to about 178, so the shard clears with margin rather than landing on the line. Every expected value was confirmed by executing the built parser before being asserted, not inferred from reading the source. --------- Co-authored-by: sim <sim@local>	2026-08-22 13:17:51 -04:00
Tom Boucher	14679b866b	enhance(#2856): add default-off live-DOM UAT capability (#3716) ... * test(#2856): add failing-first suite for the live-dom-uat capability Binds the approved triage shape before any of it exists: - containment — the execute:wave:post hook must not render unless workflow.live_dom_uat is true AND the capability resolves active (fail-closed on a missing state entry, and on a non-boolean value) - criterion 4 — agents/gsd-executor.md carries no browser MCP family; asserted as an absence, which is the only way it is observable - Hyrum guard — the pre-existing mcp__playwright__* branch must stay outside the key-gated block, or upgrading silently removes working automated UI verification for every current Playwright-MCP user - parity — the browser glob list now lives in two surfaces (agent frontmatter + workflow detection block); the assertion fails if either gains or loses a family without the other Red by construction: the capability, agent and workflow block do not exist yet. Verified on the remote runner. Refs #2856 * enhance(#2856): add default-off live-DOM UAT capability A phase whose acceptance criteria needed a live DOM could not be finished by the agent that executed it: gsd-executor carries no browser tools, so it correctly returned checkpoint:human-action even though the work was not human-only, just tool-less. Every such phase degraded to "executed, then finished by hand in the orchestrator", and autonomous: false could not distinguish "a human must judge this" from "the executor lacks the tool". Implements the shape approved at triage, not the one reported. The executor's tools: line is NOT widened, in any configuration: for a first-party agent the static list is the only control that exists (ADR-1244 D2, ADR-857 D4, no per-dispatch override). Instead one default-off capability owns the key, the agent, and the step: - capabilities/live-dom-uat/ — activationKey workflow.live_dom_uat (boolean, default false), one additive step at execute:wave:post (onError: skip, gates: []), so it can never halt a wave - agents/gsd-dom-verifier.md — the only GSD agent carrying browser MCP globs, in its own tools: line, with no Bash - verify-work automated_ui_verification — a gsd:live-dom-families block naming both new families AND the key; presence alone never activates Two independent fail-closed gates: isCapabilityActive renders a hook only on state.active === true, plus the step's own `when`. The pre-existing mcp__playwright__* branch keeps the gating it already had and stays outside the new block. Pulling it behind a default-off key would have silently removed working automated UI verification from every current Playwright-MCP user on upgrade. Also closes a host gap this surfaced: execute:wave:post dispatched only contribution + gate, so ANY registered step was declared and silently never run — exactly the single-kind hand-roll loop-hook-dispatch.md names. Step 5.75 now dispatches every kind == "step". The browser-profile lock is tolerated, not coordinated: --isolated is a flag on the operator's own MCP-server registration that GSD neither launches nor parameterizes, so the verifier reports could_not_look / profile_locked, names the flag, and stops. DOM-VERIFY.md keeps could_not_look and nothing_to_report distinct behind a closed reason enum — collapsing them is the ambiguous-run-notes defect reported. Verified on the remote runner. Closes #2856 * fix(#2856): apply review findings from the orthogonal passes Correctness pass (blocker): - delete detectionBlockIsCrlfSafe. It was pass-always: it read the file, replaced LF with CRLF, then indexOf'd marker strings that contain no newline, so the replacement could not change the result and the assertion could never fail for the reason it stated. There is no real CRLF risk on this surface either — the gsd:live-dom-families block has no parser, only human and agent readers. Deleted rather than replaced, per the repo's pass-always-test rule. Isolated security pass (two minors, both real): - execute-phase.md step 5.75: this change is what first activates kind == "step" dispatch at execute:wave:post, which newly opens the ref.command shell path at that loop point. Our own step uses ref.agent and never touches it, but the door is now open, so the step-dispatch line carries the same in-context validate-before-shell warning the sibling gate-dispatch line directly below it already carries. - gsd-dom-verifier: quoted page text in DOM-VERIFY.md is attacker influenced. Require it wrapped in inline code or a fence, kept short, and never left reading as a directive to the next reader. Verified on the remote runner. Refs #2856 * fix(#2856): settle the new-agent roster ripple Checkpoint 2 returned 28 failures, none in the new suite — all of them the guards that exist to make adding an agent a deliberate act. Each is a real boundary that had to move: - docs/AGENTS.md: Tools row must copy the frontmatter verbatim (#2526), so the browser globs lose their backticks; primary-agent counts 21->22, roster 33/34->34/35, Verifiers category 1->2 - docs/INVENTORY.md: roster completeness requires every agents/gsd-*.md to be classified exactly once - gsd-dom-verifier: add the anti-heredoc instruction and the commented hooks: frontmatter pattern both agent gates require - gsd-core/bin/shared/model-catalog.json: every shipped agent needs a profile entry (#3229) - copilot-install / kilo-upgrades / qwen-upgrades: expected agent list and the 34->35 roster boundary - execute-wave-post-gate-pipeline-e2e: execute:wave:post legitimately carries one step now. Asserted as an exact shape — one step, capId live-dom-uat, ref.agent gsd-dom-verifier, onError skip — so it stays a real guard against accidental change rather than being relaxed Two findings worth naming: mcp-tool-inheritance (#2526) rejected the agent for documenting mcp__playwright__* while its tools: line withholds it — a dead instruction that invites the agent to claim a path it cannot take. The prose now names the Playwright MCP family without the dispatchable token, in both the agent and the capability fragment. runtime-launcher-parity rejected the new gsd_run call: each fenced block is its own shell, so a workflow step file invoking gsd_run needs its own canonical preamble. Propagated with scripts/sync-runtime-launcher.cjs. That script also normalizes explore.md, which is unrelated pre-existing drift the parity check tolerates, so it is reverted to keep this diff scoped. The emitted-drift ack supersedes the spent #3370 entry for execute-phase.md — it is merged into next, so its ripple is absorbed at the base and it can no longer clear anything. That is the same supersede the #3370 entry itself performed on the spent #3324 fragment. Its unrelated execute-plan.md entry is untouched. Verified on the remote runner. Refs #2856 * fix(#2856): drop the stale emitted-drift ack entry The automated-ui-verification.md entry was written speculatively rather than from a reported growth, and the check names that precisely: an ack "written or reworded in THIS diff, but nothing here needed it, so it explains nothing". The growth tier keys on the bare filename as it appears under gsd-core/workflows/ or agents/. automated-ui-verification.md is nested under verify-work/steps/, so it was never in the tracked set — only execute-phase.md was ever reported, both before and after the launcher preamble landed. Only ack what the check actually reports. Verified on the remote runner. Refs #2856 * chore(#2856): backfill changeset pr number pr:0 -> 3716. The placeholder fails both changeset-lint (fail_invalid_fragment) and docs-lint (fail_malformed_fragment) by design and can only be resolved once the PR number exists. Both now report ok against GITHUB_BASE_REF=next. Refs #2856 --------- Co-authored-by: sim <sim@local>	2026-08-20 15:07:21 -04:00
github-actions[bot]	97ca69a53e	chore: sync next package version to 1.11.0	2026-08-19 13:51:49 +00:00
Tom Boucher	3ab0007164	enh(#2875): materialization primitives — durable user-artifact staging and descriptor-authoritative agents (#3600) ... * fix(#2875): stage user artifacts durably across install wipes (#1874-F19) preserveUserArtifacts held user files only in an in-memory Map across the wipe, so any process death between preserve and restore lost them outright. Seven call sites, not the four the issue records. Three of them never called the helper at all - they open-coded the same read/wipe/write - so searching for callers under-counted by construction; the extra sites were found by sweeping for the pattern instead. The worst is the mainline install path, where the crash window spans the entire gsd-core tree copy rather than a single rmSync. Adds src/user-artifact-staging.cts: durable on-disk staging with a record written after the copies land as the commit point, plus recovery of orphaned batches on the next run - without recovery the staged bytes survive but the user's file is still gone, which would pass its own test while delivering nothing. Routes copyPreservingSymlink through installFs() so staging cannot bypass the install fs seam, and reunites its symlink-safety docblock with the function it documents. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> * docs(#2875): amend ADR-3574 with four claims disproved by implementation Implementing Phase 6 disproved four statements the ADR rests on. The central decision - no single materializer - is unaffected and stands. Corrected: decision 3 was already satisfied, so nothing was extracted; the agents-bypass runtime set omitted claude, kilo and opencode, and closing it needed three new pieces of descriptor contract rather than proceeding on its own terms; three of the four blockers the layout comment names were already stale; and F19 is seven call sites, not four. Records the generalizable lesson: the defect is the pattern of holding user data in memory across a wipe, not the helper, so searching for callers of the helper under-counts by construction. Also resolves the ADR's open question on USER_OWNED_ARTIFACTS membership, and notes that copyPreservingSymlink needed routing through the install fs seam before it could be reused. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> * fix(#2875): close dangling-symlink blind spot and harden staging recovery An adversarial review found the F19 staging work shipped red and unsafe. Root cause, shared by two arbitrary-write findings: hasExistingSymlinkBetween missed dangling symlinks in both its root check and its per-segment walk, because it probed with existsSync, which is false for a link whose target does not exist. Fixing only the new module would have reused a guard that was itself blind. This guard protects the whole install tree. Recovery no longer throws: it degrades per entry and per file, so one bad batch cannot block the others. Previously an unrecoverable entry propagated out of the first statement of install and uninstall, before the cleanup that would have removed it - wedging the installer permanently. Partial fs adapters now throw on any omitted method instead of silently reaching the real filesystem, closing the trap that let a test poison list pass while real IO happened. Staged names must be flat, recovery refuses a dangling destination symlink, and a batch whose recovery genuinely failed is no longer swept - it was discarding the only durable copy of the file it had just failed to restore. Replaces three tests that could not fail, including the one labelled negative proof. Known limitation, documented not closed: concurrent installs sharing a staging key can still lose a batch. A real fix needs a cross-process lock. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> * enh(#2875): make the descriptor authoritative for the agents kind Deletes the inline agent-staging loop in bin/install.js and the _DESCRIPTOR_AGENTS_RUNTIMES set, so every runtime materializes agents from its capability descriptor instead of an inline hostBehaviors dispatch. Closing it needed three pieces of contract the descriptor pipeline never had, all reducible to one missing input - per-agent resolution context: a frontmatter-extensions step for claude's effort and disallowedTools, per-agent model-override resolution for kilo and opencode, and a named branding converter for hermes, whose rewrite data was already declared. Seven runtimes were on the loop, not the six the design recorded - kimi-code was found by a golden fixture, not by analysis. claude-local and kimi-code both silently lost their agents mid-change; the fixtures caught both and the cause was fixed rather than the fixtures regenerated. A parity harness gates the migration: both pipelines over identical inputs, byte-identical output including filenames, per runtime. It is demonstrated red before being trusted. Surface and install paths converge for all seven, which also fixes surface previously writing no agents for these runtimes. Codex's config.toml strip stays put - it mutates host config, which no descriptor kind models. Also routes install-model-override-resolver and install-effort-resolver through the install fs seam. Both leaked real filesystem IO from the install call tree; the stricter adapter is what exposed them. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> * docs(#2875): record the agents-descriptor migration and correct the ADR count The _DESCRIPTOR_AGENTS_RUNTIMES allow-list no longer exists, so the host integration guide told readers to join a set that is gone. Replaces that with what is now true - declare an agents entry and it installs, on the surface path as well as install - and points anyone needing a per-agent transform at the three extension points rather than at a new inline branch. Corrects the ADR amendment: seven runtimes were on the inline loop, not six. kimi-code was found by a golden fixture going red, not by reading. That is the third short count this phase, all from enumerating by symbol or set membership when the thing that matters is a behavior. Adds the Changed changeset for the surface-path convergence. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> * docs(#2875): amend ADR-2866 - claude global always wrote agents on disk The claude row's global=[skills] described what capability.json declared, not what the installer wrote. bin/install.js's inline agent-staging loop was never scope-gated and never consulted the descriptor, so a claude --global install has always written agents/gsd-*.md. Phase 6 closes the gap by deleting that loop and declaring agents on claude's descriptor at global scope. On-disk bytes are unchanged - the golden fixtures did not move, which is the evidence that the descriptor, not the installer, was incomplete. #2218 is unaffected: agents are not trigger-bearing, so the wider row does not introduce a new shadowing case. Records the warning that an incomplete descriptor is invisible while a second code path silently does its work, and only surfaces when the two are forced into agreement. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> * fix(#2875): close review findings across staging, agents and the parity harness Two independent reviews of this branch found defects the local gates missed. Security: a dangling symlink at a migration destination allowed writing outside configDir - the same class this change claimed to close, missed at the terminal write of the flow being added. The staging-root resolver threw as the first statement of install and uninstall, so a hostile symlink bricked both, and symlinked-configDir users lost uninstall as well as install; it now degrades instead of aborting. Recovery gained a source-side symlink check and now refuses a relative destDir, which resolved against cwd. Converter dispatch gained a runtime allowlist - lint-time validation stopped mattering once this branch promoted that dispatch from the surface path to real installs. Correctness: claude --local --minimal exited 1 because the minimal profile legitimately yields zero agents and the new path treated that as a failure. cline --local silently lost its agents - its descriptor declared none while the deleted loop wrote them unconditionally. The agents prune was widened to any gsd-* entry and destroyed user files it never owned. The parity harness, on which the migration's safety argument rested, drove a synthetic registry and never byte-compared the shipped descriptors; two of its trap rows could not fail. It now drives the real registry across 13 runtime-scope rows including kimi-code and cline-local, and its red-proof is demonstrated by corrupting a live capability.json. Three goldens that had encoded the cline regression as expected behavior were corrected. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> * fix(#2875): close findings from both mandated review engines /security-review found the staging source-side walk honouring GSD_ALLOW_SYMLINKED_DEST, an opt-in documented as relaxing only the write destination. A symlinked files/ component dereferenced because copyPreservingSymlink lstats the leaf only, so an intermediate link is followed. The source walk no longer honours the opt-in; the destination check still does. /code-review spec axis found this branch had reintroduced its own bug: migrateLegacyDevPreferencesToSkill's new symlink refusal threw unguarded after the legacy dir was wiped and before the staged batch was restored, so a planted symlink bricked uninstall permanently and orphaned the batch. Refusal kept, abort removed. kimi-code local silently lost its agents, the same class as the cline bug, and the parity harness recorded that exclusion as intentional - the third test in this branch to pin a regression as correct. --minimal now creates an empty agents/ dir that never existed. Behaviour restored rather than softening the changeset, so its byte-identical claim stays true. Standards axis: try/finally removed from twelve test bodies, fast-check properties added for parseOwnerPid, boundary coverage at the grace window and the ancestor-probe depth, a parity assertion for the staging-root helper duplicated across two files, and the 8-deep config walk deduplicated. Records 60-review.json with every finding and disposition from five passes, including the smells left unfixed and why. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> * fix(#2875): prune stale agents unconditionally in minimal mode The previous round stopped an empty agents/ directory being created when the resolved profile yields no agents. That was implemented by skipping the agents kind entirely, which also skipped its stale-agent prune - so a full to minimal downgrade left stale gsd-* agents behind. The deleted inline loop pruned unconditionally and only skipped writing. Those are three separate conditions, not one: prune always, write only when there is something to write, create the directory only when writing. Both call sites now run _removeGsdEntries before the empty-staged early exit. The symlink-escape guard moved with it, since the prune also touches dest. Codex .toml agents and the config.toml stanzas are cleaned again, and user-owned agents are still preserved. The agents/ directory is left in place after a prune empties it, matching every sibling kind - none of them remove the destination directory itself. Golden fixtures confirmed byte-identical: the prune is a no-op on a fresh install, so fixture generation is unaffected. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> * docs(#2875): document interrupted-install recovery for user-owned files The durable-staging fix is invisible to the user it protects. Someone whose install died mid-flight has no way to know USER-PROFILE.md was staged before the delete, that the next run restores it, or that recovery happens at the start of that run rather than in the background. Written as the task the user has - finish the interrupted command - rather than as a description of the mechanism, and states what it will not do: overwrite a file already present, or touch staging belonging to another install still running. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> * chore(#2875): backfill changeset pr number Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> * test(#2875): assert the J8 model override without building a regex CodeQL flagged incomplete string escaping: the assertion interpolated the override value into a RegExp while escaping only forward slashes, which is meaningless in a constructor, leaving real metacharacters unescaped. The failure direction was the dangerous one - a metacharacter would have made the match more permissive, so the row would pass when it should fail. That matters here because J8 exists precisely because an earlier revision was a tautology; the rewrite reintroduced a different way for the same assertion to stop discriminating. Replaced with a line-wise exact match, so no regex is constructed at all. Swept the other test files this branch adds; no sibling instances. lint:ci passed on the original - lint-no-adhoc-regex-escape matches a full metachar-escape copy, so a single slash replace slipped under it. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> --------- Co-authored-by: sim <sim@local> Co-authored-by: Claude Opus 5 <noreply@anthropic.com>	2026-08-17 17:25:53 -04:00
Tom Boucher	71180983a0	fix(#3423): standardize on <required_reading>, retire the files_to_read emit tag (#3432) ... * fix(#3423): standardize on required_reading, retire files_to_read emit tag * test(#3423): flip tag assertions, extend consistency guard to spawner surfaces * fix(#3423): sweep capabilities fragments, regen registry+skills, anchor executor test * chore(#3423): acknowledge tag-rename emitted ripples and workflow growth * chore(#3423): broaden emitted-ripple acknowledgment to all embedders * chore(#3423): settle emitted-drift acks post-rebase (merge 3004/1689-owned keys) * chore(#3423): drop stale ripple acks, ack execute-phase growth * chore(#3423): restore pristine 3004 fragment, keep only consumed appends * chore(#3423): backfill changeset pr number * chore(#3423): settle emitted-drift acks post-merge (move code-review-fix ripple into 3190, tag-rename ripples into 3191/3297) * chore(#3423): re-arm 3324 ack for execute-phase.md tag-rename ripple * fix(#3423): trim 8 bytes from execute-phase model note to hold ADR-857 margin, re-arm 3370 ack for net +4 growth --------- Co-authored-by: sim <sim@local>	2026-08-14 16:03:48 -04:00
Tom Boucher	967bddba37	fix(#3384): strip mcp__* tool grants from zcode-installed subagents (#3483) ... * fix(#3384): strip mcp__* tool grants from zcode-installed subagents ZCode's dispatcher treats every mcp__<server>__* entry in an agent's tools: frontmatter as a required MCP server and hard-fails the subagent spawn (CONFIGURATION_ERROR) when it is not connected, whereas Claude Code treats the same grants as an optional allowlist. ZCode shared Claude's verbatim agents copy (converter: null), so all 8 MCP-granted agents failed to spawn out of the box with zero MCP servers configured. Add convertClaudeAgentToZcodeAgent — a line-surgical converter that filters mcp__* entries out of the frontmatter tools: grant list (both inline comma and YAML block-list shapes) and preserves every other byte. Declare it on both of zcode's capability.json agents entries and cut zcode over to the descriptor-driven agents path (_DESCRIPTOR_AGENTS_RUNTIMES) so the legacy inline loop stops deleting+re-copying the converted agents raw. Claude Code, Kimi, and Gemini install behavior is unchanged. * chore(#3384): link changeset fragment to pr 3483 --------- Co-authored-by: sim <sim@local>	2026-08-14 11:49:01 -04:00
Tom Boucher	9dd240f634	fix(#3302): return per-agent worktree metadata from workflow scripts (#3450) ... * fix(#3302): return per-agent worktree metadata from workflow scripts * fix(#3302): link changeset fragment to pr 3450 --------- Co-authored-by: sim <sim@local>	2026-08-14 09:28:22 -04:00
0xdhx	0396d9cab1	enhance(#2483): stop the claude reviewer lane from inheriting CLAUDE.md + auto-memory (#2493) ... * enhance(#2483): env-guard the claude reviewer leg against CLAUDE.md injection The claude reviewer in workflows/review.md was a bare headless `claude -p` spawn run from the project cwd, so it inherited the invoking user's global CLAUDE.md, the project CLAUDE.md, and Claude Code auto-memory. That made it the only reviewer leg seeing anything beyond the prompt file. gather_context assembles PROJECT.md, the roadmap section, every PLAN file, CONTEXT.md, RESEARCH.md and REQUIREMENTS.md into the prompt before any reviewer runs; the gemini leg receives only that prompt and the codex leg runs --ephemeral. Beyond the measured ~4k tokens/spawn, the asymmetry cuts at the workflow's own premise: "independent review" meant something different for the claude leg than for the other two. Guard both dispatch lines with a per-invocation `env CLAUDE_CODE_DISABLE_CLAUDE_MDS=1`. `env`, never `export` — the flag must not leak into the orchestrating session (which may itself be Claude Code on the SELF_CLI="auto" path) or into any later spawn. review.md is the only claude -p call site in the installed tree, so this is two lines on one surface. The self-skip logic is untouched. * enhance(#2483): fix CRLF-fragile split and regenerate workflow baselines Two CI failures from the first push, both mine: 1. lint-tests: the new regression test split readFileSync content on a literal "\n". On a Windows git-autocrlf checkout that leaves a trailing "\r" on every line (local/no-crlf-fragile-split). Use .split(/\r?\n/). 2. golden-install-parity / workflow-size-budget / workflow-compat: editing gsd-core/workflows/review.md changes its content hash and byte size, and both are pinned in committed baselines. Regenerated via the repo's own generators (npm run size:baseline, npm run gen:golden). The regenerated diffs are review.md-only: exactly one hash line per golden-install-parity fixture and one size entry in workflow-size-baseline — no unrelated drift swept in. Full suite now green locally: 2113 pass, 0 fail, 3 skipped (run with HOME and CLAUDE_CONFIG_DIR overridden to throwaway dirs; live profile verified untouched afterward). * enhance(#2483): adapt guard-test matcher to the effort-args dispatch reshape The effortSurface wiring (#2481) reshaped the bare-model dispatch to `claude $CLAUDE_EFFORT_ARGS -p -`; the invocation matcher's dash-first form could no longer see it, and the count assertion failed exactly as designed. The matcher now tolerates variable expansions between `claude` and its first literal flag. Negative-controlled both ways: a stripped guard and a deleted dispatch line each still fail. * enhance(#2483): also guard the claude leg against auto-memory injection CLAUDE_CODE_DISABLE_CLAUDE_MDS suppresses CLAUDE.md file loading; auto-memory is an independently-toggled mechanism with its own flag. Add CLAUDE_CODE_DISABLE_AUTO_MEMORY=1 to both dispatch lines, correct the docs/COMMANDS.md and changeset claims that credited the first flag with covering auto-memory, and extend the regression test to require both flags on every claude invocation (negative-controlled: 2/4 assertions fail with the new flag removed). * enhance(#2483): match the claude binary in command position, not argument position The line-oriented invocation matcher counted any line where the token `claude` was followed by a flag. #2589 (landed on next as 920a5f3f) reshaped the effort-args lookup from --host claude 2>/dev/null | jq -r '.effort_argv_string // ""' to --host claude --pick effort_argv_string which put a flag immediately after `claude` and made the config query read as a third claude dispatch, failing the count assertion. The defect class is a binary name in *argument* position being read as a command. Fixed at the class rather than the instance: tokenise the line and skip any `claude` whose preceding token is a flag. That also covers the latent sibling one line away in review.md (`command -v claude`), which escaped today only because its next token is a redirect. Negative-controlled four ways: stripping CLAUDE_CODE_DISABLE_AUTO_MEMORY=1 fails, stripping the whole env guard fails, adding a genuine third unguarded dispatch (`timeout 900 claude --output-format text -p -`) still fails — so the narrowing did not blind the matcher to reshapes, which is the property the count assertion exists for — and the pre-#2589 jq form of the lookup still passes, so the matcher is not pinned to today's base. * enhance(#2483): carry the claude reviewer's memory guard as declared lane data ADR-2782 Phase 5b replaced the hand-authored per-CLI dispatch legs in review.md with the declared lane table, so the two `env`-prefixed shell lines this PR previously added no longer have a surface to live on. The guard is reimplemented where the lane contract now lives. `SpawnInvoke` gains an optional `env`, the claude lane declares the pair, the resolver folds own string-valued entries into `SpawnPlan.env` (absent or empty resolves to null, so the runner has one shape to test), and the runner passes it to spawn. Production merges it OVER `process.env` into a fresh object for that one child, so nothing reaches the orchestrating session or any other lane in the run. Declared data rather than a handler (D6): the pairs are static per lane, which is precisely what the manifest vocabulary is for. The capability manifest carries the same field, because the lane-fidelity test compares manifest and descriptor over the union of `invoke`'s keys. The regression test is rewritten against the resolver and runner rather than review.md's text. It gains the property the source-text assertions could only approximate: that `process.env` is never mutated. Scope boundary, asserted rather than left in prose: `env` is not part of the trust-disclosure surface, which is safe only while no manifest body reaches the resolver — the registry's reviewer bodies contribute slugs to the parity check and execution resolves from `REVIEWER_LANES`. The new test fails first if that ever changes. * enhance(#2483): restate the guard's mechanism in the docs and changeset Both described the fix as two `env`-prefixed dispatch lines, which is the surface ADR-2782 Phase 5b removed. The user-visible behaviour is unchanged; the carrier is not, and a changeset that ships a description of a mechanism the tree does not have is a CHANGELOG entry nobody can verify against the code. * enhance(#2483): cover the production spawn wiring end to end The unit tests stop at the runner's `deps.spawn` seam — every one injects a spy. Production supplies that seam in `gsd-core/bin/gsd-tools.cjs` as a hand-written object no test constructs, so the chain could be correct all the way to `SpawnPlan.env` and the merge could still be wrong or absent with the suite green. Deleting those four lines was the one mutation that left every other control silent. This runs the real `spawnSync` through `gsd-tools review-lane invoke`, with a `claude` shim on PATH that records the environment it was handed. It asserts both halves in one test: the pair arrives, and an unrelated inherited variable survives — a wiring that REPLACED the environment rather than merging over it would satisfy the first and break every lane's PATH and HOME. POSIX-only; mediating a Windows `.cmd` shim is a separate concern the repo already tests on its own. Noted rather than fixed: `timeout`, `killSignal`, `maxBuffer` and `shell: false` on that same object are equally uncovered. That is the epic's gap, not this change's, and closing it is not in scope here. * enhance(#2483): validate the invoke.env shape and register it as spawn-only `env` was the one spawn-invoke field with no shape enforcement: every sibling in `validateSpawnInvoke` is checked, and a manifest declaring `env` as an array, a string, a number, or an object with non-string values passed validation in silence. That matters more than an ordinary schema gap here, because `resolveLanePlan` DROPS a non-string value rather than coercing it — so an unvalidated manifest declares a pair that never reaches the spawn, which is the failure a memory guard can least afford. Two registrations, not one. `env` was also absent from `SPAWN_ONLY_INVOKE_FIELDS`, which is the list the openai-http arm rejects against — so `invoke.env` was accepted on a transport that issues an HTTP POST and has no child environment at all. It was the only spawn-shaped field accepted there; the other six each produce two errors. Self-found while sweeping the class, not raised in review. Keys are held to the portable POSIX environment-name grammar. That is a policy, not a claim about what an environment can hold: measured, only NUL is actually rejected by `spawnSync`, while `=`, a leading digit, a dash and a space are all carried through to the child (an `A=B` key arrives as the raw entry `A=B=value`). They are refused because a name outside the grammar is not portably addressable by the program meant to read it. `__proto__` is refused for a different and concrete reason. It passes that grammar and is a real own key once a manifest is JSON-parsed, but assigning it onto a plain accumulator goes through the inherited `__proto__` setter rather than creating an own property — and for the string values this field permits the setter is a no-op that does not even change the prototype. The pair would validate and then simply vanish before the spawn. (An environment CAN carry a literal `__proto__` entry; this is about the resolver's accumulator, and the error message says so.) Deliberately narrower than the sibling reserved-name guards in this file, which also reject `constructor`/`prototype`: those guard bracket lookups that resolve prototype members, whereas this reads via `Object.keys` plus an own-value read, where `constructor` assigns as an ordinary key the spawn could carry. `effortChannel` is deliberately left in neither field list: ADR-2782 D2 defines it for both transports, so it is shared rather than spawn-only. Reversion-controlled, three mutations, all three fire a named test: dropping `env` from the discriminator fails `httpTransportRejectsEnv`; removing the `__proto__` arm fails `envRejectsProtoKeyThatWouldSilentlyVanish`; disabling the block fails four. (#2483) * enhance(#2483): amend ADR-2782 D2 for the invoke.env vocabulary widening D2 records the spawn `invoke` shape as a closed vocabulary, and its Amendments section carries a dated entry for every prior widening (Phase 1 #2794, Phase 2 corrections #2795, Phase 5b #2799). This change extended that vocabulary in code without touching the ADR governing it, so the ADR contradicted the implementation — and the repo's own convention, recorded in CONTEXT.md, is that the ADR is amended in the same PR precisely because the prior widenings did it correctly. Adds the `invoke.env` row to the D2 table and a dated Amendments entry. The entry also corrects the authority this change cited. The source comment pointed at D6, which governs the closed `handler` enum — imperative behavior admitted first-party — and says nothing about the `invoke` field vocabulary. That is D2's territory, so the citation never covered the gap. Two claims are corrected rather than restated, both about the trust boundary that justifies leaving `env` out of the D5 disclosure signature: - The regression test does not enforce that boundary. On one forged lane it shows the resolver folds whatever it is handed, so a future path feeding it manifest lanes would not make any assertion in that test fail. Its comment claimed it "will fail first"; that was wrong, and both the comment and the ADR now say the boundary is a property of the production call chain instead. - The ADR is internally inconsistent on whether third-party manifest lanes execute at all: Consequences says adding a reviewer needs "no core patch", while `gsd-tools.cjs` rejects every slug absent from the first-party REVIEWER_LANES map. CONTEXT.md, `workflows/review.md` and the resolver's own header take the first view. #2483 did not create that inconsistency and does not resolve it; the entry records it rather than settling it in its own favour. (#2483) * enhance(#2483): document invoke.env in the capability-manifest reference ADR-2782 points capability and plugin authors at `docs/reference/capability-manifest.md` as where the lane vocabulary must be visible, and its `invoke` row enumerates the spawn sub-shape field by field. `env` was absent from that table while being part of the real shape, so the one document a third-party capability author would actually consult to learn the field exists did not mention it. Squarely Diataxis reference material — a field-by-field schema description — so it goes here rather than in the user-facing prose, which was already updated. States the constraints a manifest author can actually trip, and is explicit that the name grammar is a portability policy rather than an OS limit, so a reader does not take it for a claim about what an environment can hold. (#2483) * enhance(#2483): disclose and sign the reviewer lane's env and residual invoke fields `invoke.env` was undisclosed at install time. That was defensible while manifest lanes could not execute — the premise this PR's own ADR amendment recorded — and #2927/#3062 retired it: `routeReviewLane` now merges installed overlay `reviewer` bodies into its lane map via `mergeReviewerLanes`, which is a field-identical merge by ADR-2782 D1 and deliberately does not deep-validate. An overlay's whole `invoke` therefore reaches `resolveLanePlan`, and `env` reaches the spawned child. A consented third-party capability could set `NODE_OPTIONS=--require ./evil.js` on a reviewer lane with no install-time disclosure and no re-consent. The same file already decided what `env` means in a manifest: MCP servers fold it into the disclosure signature and render each key and value in the consent prompt, with an inline rationale naming this exact shape. Reviewer lanes get the identical treatment. `env` was the ninth unsigned invoke field, not the first. `defaultHost` (the manifest's OWN fallback egress host, used whenever the config key resolves to nothing), `path`, `outputChannel`/`outputArg`, `modelArg`, `effortChannel` and `modelDiscovery` all reach `resolveLanePlan` and none was bound. Enumerating a ninth name leaves the tenth open, so the lane signature carries a RESIDUAL of every other declared `invoke` key — the completeness backstop `rawConfig` already gives the MCP line (#1459 finding 5), and the "sign the whole object" remedy the recorded decision on this class prefers. `defaultHost` is also rendered: `resolvedHost` comes from user config, so a lane whose key is unset displayed "(unresolved …)" — which reads as "no destination" — while the runtime egresses the plan and review text to the address the manifest picked. D4.5 is preserved one level down: the extra element is appended ONLY when the lane declares something beyond the eight already-bound fields, so an env-free lane's signature stays byte-identical and no already-consented capability is re-prompted for a field it does not use. A lane that does declare one re-consents, which is the point. Execution-primitive env names are FLAGGED in the prompt, not refused in the validator. A denylist cannot be the boundary here: `PATH` alone is a complete execution primitive for a spawn lane and can never be refused, the child is an arbitrary third-party binary so the true set spans every interpreter's injection vars, and the MCP `env` this mirrors refuses nothing and discloses everything. Missing a name costs a quieter line, never a boundary. Refs #2483. * enhance(#2483): exercise the real overlay merge path in the guard test The test named for the manifest/first-party boundary did not test it. It built a forged lane locally, handed it straight to `resolveLanePlan`, and asserted that `REVIEWER_LANES` did not contain it — so no assertion in it depended on the claim its name made, and a code path that fed manifest lanes to the resolver would not have made it fail. Its own comment said as much, and named the production chain as the real carrier of the guarantee: "gsd-tools.cjs builds its lane map solely from REVIEWER_LANES". That sentence is now false. #3062 merged overlay reviewer bodies into that map, so the test's premise and its subject both moved. The replacement routes through `mergeReviewerLanes` — the real helper the production path calls — and asserts the overlay lane is admitted, resolves, and carries its `env` into `SpawnPlan.env`. That makes the security property falsifiable instead of narrated. It then asserts what now backs it: the env is disclosed on the surface, rendered key and value in the consent prompt, flagged when the name is an execution primitive, and bound to the signature so a value change, an addition, or a removal each force re-consent. Three further cases, because the finding's generative half is what stops it recurring: the residual backstop is asserted against five fields including one that does not exist (`aFieldThatDoesNotExistYet`), so a future vocabulary widening cannot silently re-open this; a fully-enumerated lane is pinned to its original 8-tuple, which is what keeps the fix from re-prompting every consented capability; and an http lane's manifest-declared `defaultHost` is asserted to reach both the prompt and the signature. Reversion-controlled, seven mutations, all seven fail a named test: env dropped from the surface, the prompt's env line removed, the execution-primitive warning removed, the signature's extra element never appended, the residual emptied, the defaultHost line removed, and the declares-something test un-widened. The last of those was SILENT on its first run and its test was written in response, then the control re-run. Refs #2483. * enhance(#2483): correct the ADR amendment's manifest-lane premise The amendment argued `env` needed no D5 disclosure because a manifest's `invoke` fields never reach `resolveLanePlan`. That was true when written and #3062 retired it 22 hours after this branch's last commit: `routeReviewLane` now builds its lane map from `mergeReviewerLanes(REVIEWER_LANES, loadRegistry({includeInstalled: true}))`, and D1's no-translation-layer rule makes that a field-identical merge, so an overlay's whole `invoke` reaches the resolver and executes. The entry had named this exact trigger — "were manifest lanes ever made executable, `env` must join the disclosed surface in that change, and nothing here will trip if it does not." Nothing tripped. The premise is rewritten to current truth rather than annotated, because an ADR is read in fragments and a superseded paragraph left standing reads as live reasoning to the next author; a one-line dated tombstone points at git for the withdrawn text. The rewritten entry records four things the first draft could not: that the enumeration itself was the defect (`env` was the ninth unbound `invoke` field, and `defaultHost` and `path` are egress-relevant on their own), that the residual is what closes the class, that D4.5's byte-identical-signature property is preserved by appending the residual only when a lane declares something beyond the eight bound fields, and that consent — not shape validation — is the boundary, since no honest env denylist can exclude `PATH`. It also closes the internal inconsistency the previous entry could only record. This ADR, `CONTEXT.md`, `gsd-core/workflows/review.md` and `resolveLanePlan`'s own header all said overlay lanes reach the resolver while the runtime said otherwise; #3062 resolved that in the documents' favour, which is what makes the disclosure mandatory rather than defensive. Refs #2483. * enhance(#2483): record in the manifest reference that invoke fields are consent-bound `docs/reference/capability-manifest.md` is the field table ADR-2782 points capability authors at, and it described `invoke` purely as a schema. A third-party author reading it could not learn that everything they declare there is shown to the user at install and bound to the consent signature — which is exactly what they need to know now that an overlay reviewer lane executes (#2927/#3062). States the two things the schema alone cannot: that `env` and `defaultHost` are named in the consent prompt and the rest is covered by a residual, so any change to a declared `invoke` field forces re-consent; and that `env`'s validation is a portability policy rather than a safety boundary, since `PATH` is a complete execution primitive and cannot be refused. Names that are execution primitives are highlighted in the prompt instead. Refs #2483. * enhance(#2483): add a Security changeset for the reviewer-lane disclosure The existing fragment describes the enhancement this PR was opened for and stays as it is. The disclosure fix is a separate user-visible change of a different type: a capability declaring `invoke.env` or `invoke.defaultHost` will ask for consent once more, and users are entitled to read why in the changelog rather than discover it as an unexplained prompt. Type is `Security` rather than `Changed` because the entry describes a closed code-execution disclosure gap, not a behaviour adjustment. Refs #2483. * enhance(#2483): correct this round's own claim about who gets re-prompted Self-found while auditing the round's claims before publishing them. The changeset and the ADR entry both stated that a capability declaring `invoke.env` or `defaultHost` "will ask for consent once more". That is wrong, and it overstated the cost of the fix in the one direction a maintainer would have had to take on trust. A code change to `disclosureSignature` re-prompts nobody. `hasProjectConsent` matches on the recomputed bundle `contentHash` — the signature has not been the security binding since #1459 CB-1/CB-2 — and the upgrade path's `executableSetChanged(old, new)` compares two disclosures both computed by the CURRENT code, so widening the signature moves both sides of that comparison equally. First-party capabilities never reach the path at all: the install flow blocks a first-party id before trust evaluation. What the widening actually buys is forward-looking, and is the real argument for it: an upgrade whose manifest edits a declared `invoke` field now registers as an executable-surface change and re-consents, where before it could change what the lane runs in silence. Also measured and recorded, because the D4.5 property was stated more strongly than it deserved: of the twelve first-party reviewer capabilities, ZERO are in the byte-identical-signature class — every real lane declares at least `effortChannel`. The property is a guarantee about minimal lanes, not a description of the fleet, and the ADR now says so. Refs #2483. * enhance(#2483): sign and disclose the probe binary and the lane's outer fields Found by this round's own adversarial review, and it is the same defect one level out: the `invoke` residual cannot reach the lane body's OUTER fields, and `probeLane` SPAWNS `probe.binary` with `--help` before dispatch (`review-lane-runner.cts`, the `command-exists`/`command-capability` arms). An overlay naming an arbitrary probe binary therefore executes it — unsigned and undisclosed, exactly as `invoke.env` was, and reachable on the same #3062 path. The lane element now carries a second residual over the outer fields, and the probe binary is shown in the consent prompt when it differs from the dispatch binary — it is a program that runs, and the user is entitled to see it. TWO fields stay excluded, and that is a decision rather than an omission: `reviewsSection` and `timeoutFloorMs` are ADR-2782's cosmetic carve-outs (matrix A10/A13), where re-consenting would present a prompt carrying no security information. A test pins that they remain excluded, so a later widening cannot quietly reverse D4.5 while claiming to complete this fix. Also corrects a miscount introduced by the previous commit: the source comment said the enumeration had fallen behind by "seven fields" and omitted `fallbackModel`, while asserting `env` was the ninth. `resolveLanePlan` reads twelve `inv.*` fields and four were bound, so the number is eight. The comment now states the derivation rather than just the total. Reversion-controlled: emptying the outer residual fails "repointing the probe binary must force re-consent"; removing the render line fails its own named assertion. Refs #2483. * enhance(#2483): refuse execution-primitive env names as defence in depth Adopts the review's B5 after this round's own adversarial pass refuted my reason for declining it. I had argued a denylist was worthless because `PATH` can never be refused. That was wrong on the facts: no shipped reviewer manifest declares `PATH`, so it can be refused, and it is the most complete primitive in the set — repoint it at a directory holding a fake binary and the declared `invoke.binary` is irrelevant. A list that cannot be exhaustive can still close the highest-confidence, lowest-legitimacy routes. So the validator now rejects `PATH`, `NODE_OPTIONS`, `LD_PRELOAD`, `DYLD_INSERT_LIBRARIES`, `BASH_ENV`, `PYTHONPATH`, `PERL5OPT`, `RUBYOPT`, `GIT_SSH_COMMAND`, `JAVA_TOOL_OPTIONS` and their siblings on a reviewer lane. A lane needing a specific executable declares an absolute `invoke.binary` instead of reshaping the child's environment. The comment states plainly that this is defence in depth and NOT the boundary — the boundary is install-time consent, which discloses every declared pair and binds it to the signature, so an unlisted name is still SEEN before it runs. That framing is load-bearing: a future reader who mistakes the denylist for the control will under-invest in the one that is, which is the failure mode I was trying to avoid by declining it outright. Two tests: the rejection itself across ten names, and a guard asserting no shipped reviewer capability declares a denied key — so if the list ever outgrows its evidence, that surfaces as a decision rather than a silent removal. Refs #2483. * enhance(#2483): fix two stale D5 enumerations elsewhere in the ADR The previous commit rewrote the amendment's premise but swept only the amendment. Two normative passages earlier in the same ADR still enumerated the old closed field list and now contradicted it: the `executableSetChanged` trigger list, and the split-binding note asserting the seven manifest-derived fields were "everything that is SHA-pinned". That is the failure the rewrite-don't-annotate rule exists to prevent, one section over — an ADR is read in fragments, and a fragment carries no supersession marker, so a reader landing on either passage would have taken the superseded enumeration as current. Both now name the residual as the mechanism rather than restating a list, which is also what stops them going stale the next time the vocabulary widens. Found by this round's adversarial review, which grepped the whole document rather than the section under edit. Refs #2483. * enhance(#2483): stop the probe disclosure claiming a spawn that does not happen The probe line added one commit ago rendered "probes by running: <binary> --help" for every lane. That is false for `kind: "command-exists"`, which only calls `hasBinary` — a PATH/filesystem scan that starts no process. Only `command-capability` spawns. A false statement in a consent prompt is worse than a missing one: the prompt is the surface a user is asked to trust, and this one overstated what a lane does. Worse, the test I wrote to prove the fix used `command-exists` — the kind that does NOT spawn — so it pinned the wrong claim and would have kept the error green forever. The surface now carries `probeKind` and the two kinds render differently: a spawn is described as a spawn, a presence check as a presence check. The test exercises both, and asserts the `command-exists` path never emits the spawn wording. Also corrects the field-count parenthetical to state its derivation unambiguously — `resolveLanePlan` reads thirteen `inv.*` fields including `env` (twelve before this PR), four were bound, so eight were unbound before `env` and nine including it. The bare "twelve" was true only of the pre-PR tree and read as a claim about the current one. And retires two comments that argued AGAINST the validator denylist this round then shipped. Leaving them would have handed the next reader the reasoning for removing it. Reversion-controlled: conflating the two probe kinds fails a named test. Refs #2483. * enhance(#2483): match the reviewer-lane env denylist case-insensitively The denylist added one commit ago compared exact case, so `Path`, `path`, `node_options` and `Node_Options` all passed it. Windows environment lookup is case-insensitive, so those reach the child as `PATH` and `NODE_OPTIONS` — the exact inputs the list names. An exactly-cased denylist is worse than none: it reads as a control while admitting the input it was written to refuse, and the next reader has no reason to doubt it. Members are stored uppercase and the key is folded before lookup; the name grammar already constrains keys to ASCII, so a plain fold is sufficient. Reversion-controlled: restoring the exact-case compare fails `envDenylistIsCaseInsensitive` on `Path`. Refs #2483. * enhance(#2483): correct the docs that still described the denylist as absent Both the ADR and the manifest reference still said `env` carries no denylist and that `PATH` "can never be refused" — written when that was this round's position, and left standing after the round reversed it. A reader landing on either passage would have taken the superseded argument as current, which is precisely the failure the rewrite-don't- annotate rule exists to prevent. Both now describe the denylist, name `PATH`'s inclusion and the case-insensitive match, and keep the limit explicit: the list cannot be complete against an arbitrary child and disclosure runs before validation, so consent remains the boundary. The ADR's byte-identical-signature claim is also corrected rather than softened. With the outer residual in place, a lane producing no residual is one the validator rejects — it declares no `flags`, `probe`, `emptyOutput`, `evidenceClass`, `requiresBinaries` or `promptBudgetKey`. So the property is about the ENCODING, not a claim that any real signature is unchanged, and it is not the argument for the change being safe. That argument is that consent binds to the bundle contentHash and no existing consent is invalidated at all. Refs #2483. * test(#2483): cover the three new lane disclosure fields in the injection-safety parity guard The PARITY test in section N exists to catch a renderer field that skips `renderValueForPrompt` (#3248). Its payload manifest is hand-maintained, so it covers the fields that existed when it was written — slug, binary, args, hostConfigKey, handler — and none of the fields this PR adds. This PR renders three further manifest-supplied values into consent-prompt lines: `invoke.env` (keys and values), `invoke.defaultHost` and `probe.binary`. The gap was silent rather than theoretical: with the lane env line reverted to the pre-#3248 raw form, the whole 948-test lane/capability/trust-disclosure suite stayed green. Two manifests, because the shapes render disjoint lines — `defaultHost` only on the openai-http branch, `env`/`probe` only where declared, and the probe line only when the probe binary differs from the dispatch binary. Non-vacuity is asserted on the typed disclosure object and on structural line counts, not by substring-matching rendered prose: CONTRIBUTING.md forbids raw text matching on test output, and this section's own header promises structural assertions only, so a prose match here would have made that promise false. Negative-controlled three ways against the merged tree, each producing exactly one named failure: env rendered raw, defaultHost rendered raw, probe binary rendered raw. * docs(#2483): extend the #3248 render-site comment to the fields this PR adds The comment enumerates every manifest-supplied value that must pass through `renderValueForPrompt`, and it stopped at `handler` — the reviewer-lane fields that existed when #3248 landed. This PR renders three more (`defaultHost`, the probe binary, and the env keys and values), so the list understated its own contract in the one place a future author would check before adding a fourth. A comment enumerating a closed set is a set that can silently fall behind the code it describes; the parity test added alongside is what makes the omission fail loudly rather than read as deliberate. --------- Co-authored-by: Tom Boucher <trekkie@nomorestars.com>	2026-08-11 17:42:41 -04:00
Tom Boucher	cf6de5e1c0	feat(#2871): resolve triggers and host precedence, not just placement (#3291) ... * test(#2871): failing-first suite for trigger-surface resolution 23 tests over the 50-test-matrix rows. RED by construction: resolveTriggerSurface and DEFAULT_TRIGGER_PRECEDENCE do not exist yet, and the validator silently ignores triggerPrecedence today. Written in the per-runtime describe idiom the other four runtime-artifact-layout suites use, not a table. The rows that carry the weight: windsurf must NOT report a shadow it does not have, since its global scope emits only agents and agents are not trigger-bearing; agents and kimi-agents must be absent from the output for every runtime; and reordering a runtime's triggerPrecedence must flip the winner, which is the only assertion that proves the axis is read rather than decorative. Stems are injected, never scanned, so the surface is assertable with no filesystem. * feat(#2871): resolve triggers and host precedence, not just placement resolveTriggerSurface(runtime, scopes) returns every /gsd-<name> trigger a runtime emits, with the scope and kind that produced it, whether the host registers it directly or only through a router, and which artifact shadows it. resolveRuntimeArtifactLayout is untouched -- its 7 callers need placement only and the issue requires them unchanged. AGENTS ARE NOT TRIGGER-BEARING, and ADR-2866 said they were. The host-integration matrix models command and dispatch as separate interface points: an agent is invoked through the Agent tool's subagent_type, not by typing a slash trigger, and _copyStaged never applies the kind prefix to an agents entry. So agents and kimi-agents are excluded from the surface entirely, and this commit amends ADR-2866 with a dated correction. #2218's conclusion is unchanged -- the collision is strictly commands-vs-skills, and claude's local /gsd-* trigger surface is still fully shadowed -- but the ADR implied the local agents surface was lost too, and it is not. That correction is what makes windsurf come out right. Its global scope emits only agents, so it has no global trigger and its local commands are unshadowed. Model agents as trigger-bearing and windsurf falsely reports a full shadow. The triggerPrecedence axis lands on all 19 descriptors as an ordered kind list, one value with one owner, rather than a numeric rank spread across N kind entries with nothing keeping them consistent. Validation uses a required-with-default shape that has no precedent in this validator -- every existing axis is hard-required -- so a third-party capability.json omitting the field still validates, which is what ADR-894's additive-only contract promises. Winner resolution reads Phase 1's scope rank first, then the kind ordering. A test reorders the axis and asserts the winner flips, since an axis that is added, validated and never consulted would pass every other assertion. shadowedBy ships unread. Phase 4 (#2873) is its first consumer, per this issue's out-of-scope note. Verified via the remote runner. * fix(#2871): single-source namespacedByDir and close two test gaps Four findings from the isolated adversarial review. The namespacedByDir rule had reached three copies -- install-engine, surface, and the new trigger resolver -- one of which carried a hand-written keep-in-sync comment and no assertion. That is this repo's generative-fix-divergence class. Extracted to one exported predicate all three now call. Verified by diverging one copy deliberately: the existing #816 parity test failed, and passes again on revert. The omission test was vacuous. Row 16 asserted that a descriptor without triggerPrecedence still validates, but built its fixture from claude's shipped descriptor -- which this PR had just added the axis to. It now clones and deletes the key, following the shippedDescriptorWithout pattern, and asserts both that validation passes and that the resolver still picks the right winner from the default. The second half is what makes it prove anything. resolveTriggerSurface silently dropped an unrecognized scope while every sibling in this epic throws. Two phases of one epic should not disagree about whether an invalid scope is an error, so it now rejects through the same shared validator; an empty scope list still returns empty rather than throwing. The ADR amendment had been spliced into the middle of the References list, orphaning its last bullet. Moved to the top, after the header block, which is where ADR-3660 and ADR-1016 both put dated amendments. No lint checks markdown structure, so this was green while malformed. * fix(#2871): single-source the command filename composition too The earlier fix shared the namespacedByDir boolean but left the filename composition around it written twice -- once in _copyStaged as what actually gets written, once in resolveTriggerSurface as what gets predicted. The predictor could go stale silently. One exported helper now composes it for both. The entry.name asymmetry that looked like it would block extraction does not: entry.name is filtered to end in .md and stem is entry.name minus those three characters, so the two branches are the same string by construction. Divergence proven to fail: injecting a marker into the helper broke the trigger-surface suite; reverting restored 25/25. The four sibling layout suites hold at 227 unchanged. * docs(#2871): correct the ADR timing notes that this phase makes stale The Amended by back-links on ADR-3660 and ADR-1016 were written in Phase 0, when the widenings they describe had not shipped. Each carried a forward-looking clause -- "the module changes at Phase 2, not before, until then this module resolves placement only" -- which becomes false the moment this PR merges. ADR-2866's own Amends header and its reciprocal-notes section carried the same tense. All four now describe what shipped. This is a tense and status correction on Accepted ADRs, not a change to any decision. Worth stating because it is the failure mode this epic keeps meeting: gen-adr-index.cjs tracks only Supersedes and Subsumes, so nothing in CI would have caught either the missing back-link in Phase 0 or these stale clauses now. They stay correct only because someone checks. * chore(#2871): backfill changeset PR number --------- Co-authored-by: sim <sim@local>	2026-08-09 22:25:42 -04:00
Tom Boucher	b901d1e06f	feat(#1953): complexity-triggered refactor extension point (execute:post) (#3261) ... * test(#1953): failing-first suite for the complexity-triggered refactor hook 60 behavioral cases against src/complexity-trigger.cts, which does not exist yet: decision-point counting, the comment/literal stripping leak surface, threshold and jump-delta boundaries at limit-1/limit/limit+1, stable-anchor baseline semantics, and fs fault injection via mock.method. Two fast-check properties assert that stripping never manufactures a decision point and that comments and string literals are score-neutral. Also registers the refactor-trigger capability manifest (inert until refactor.trigger_enabled) and regenerates the capability registry and matrix. Verified RED on the remote runner before any implementation exists. * feat(#1953): complexity-triggered refactor extension point Adds the opt-in refactor-trigger capability. After a phase executes, an execute:post step measures per-function complexity for the files the phase touched and writes a scoped refactor proposal when a function crosses the configured threshold or drifts past its recorded anchor. Design notes worth carrying: - The signal is computed in-core (decision-point counting over comment- and literal-stripped source, Node builtins only) rather than via Memtrace or a shelled-out analyzer. The hook fires as a deterministic CLI, not an agent with MCP tools, and core takes no external dependencies — this is the only option a behavioral test can bind to. The metric sits behind a seam. - The baseline is a stable anchor, not a rolling value: set on first observation, moved only on disposition. A rolling baseline makes the delta the single-phase change, so a function creeping +2 per phase never trips a delta of 5 and the jump check adds nothing over the absolute threshold. - Strict mode records an open deviation window in the broken-windows ledger rather than declaring its own ship:pre gate. ship.md has no generic ship:pre gate dispatch — only two hardcoded branches — so a third gate of any kind would be declared and never evaluated. - The gate clears on the proposal being dispositioned, never on the score improving. A blocking complexity number is one an executor can satisfy by splitting a coherent function in two. execute-phase.md gains a generic execute:post step-dispatch contract; it previously matched only ref.skill == "code-review", so any other step registered there was declared and never run. The code-review branch is unchanged. Full rationale in ADR-1953. Closes #1953 * fix(#1953): close git option injection and symlink escape in the refactor hook Three findings from the isolated security review, all fixed inline. HIGH — changedFilesSince interpolated the --since value into a revision token placed before the -- separator. A -- only stops PATHSPEC parsing of arguments after it; git still option-parses what comes before. So --since '--output=/tmp/x' became --output=/tmp/x..HEAD, which git accepts as --output=<file> and uses to redirect diff output — an arbitrary write. Fixed with --end-of-options before the revision range plus a conservative ref validator. The validator deliberately permits ~ ^ @ { } because those are legitimate git REVISION syntax (HEAD~1, main@{yesterday}) as distinct from ref-NAME syntax; --end-of-options is the actual barrier. The doc comment asserting the trailing -- was sufficient was wrong and is corrected. MEDIUM — resolveConfinedPath confined by string prefix only, so a symlink committed inside the repo passed the check (its own path is under cwd) and readFileSync then followed it outside the root. Now lstat-checks for a regular file and skips anything else with REFACTOR_FILE_UNREADABLE, so one bad path skips one file and the run continues. LOW — the new execute:post dispatch contract showed the gsd_run example before the rule requiring ref.command be validated first. That prose is executed by an agent, so textual order is execution order. Reordered. Refs #1953 * fix(#1953): make the analyzer able to see TypeScript at all Found by running the shipped analyzer over its own source: it reported functions=1 for a 940-line module with 24 function forms. A return-type annotation or a generic parameter list made a function invisible — `function f(a): number {}` and `function f<T>(a: T): T {}` both detected as zero. Since gsd-core is written in .cts and the capability declares .ts/.cts/.mts analyzable, the feature silently found nothing in this repo's own primary language while reporting success. A safety net that reports "all clear" because it cannot see is worse than no safety net. All 98 tests passed over this, because every fixture was plain JS — the exact failure the test matrix's own "assert against the shape production uses" warning describes. Adds a TypeScript-shapes suite covering return types (including unions, generics, object literals and type predicates), generic parameter lists (constrained and defaulted), export/async/generator combinations, annotated arrows, class-method modifiers, and optional/ default/rest params — plus the two traps: an overload signature has no body and must not count, and `a < b && c > d` is a comparison, not a generic. Detection now reports 24/37/21 functions for the three source files, which matches a hand count exactly. Also from review: - The strict-mode ledger dedup identified entries by parsing a prose description string. That is banned by CONTRIBUTING's raw-text-matching rule and was a real bug: the "exactly one window per untriaged proposal" guarantee rested on prose matching, so rewording a description or editing WINDOWS.md by hand silently produced duplicates. Now matches structurally on kind + phase + file + line. - A property test asserted on the stripper's output text. Reframed to assert the same invariant through analyzeSource's score. - nextBaseline's `candidates` parameter has been dead since the anchor change; removed from the signature and all call sites. - Extracted the duplicated require-or-degrade and capability-check boilerplate. - ADR-1953's Implementation bullet still named a `refactor.ship-gate` in check-command-router.cts — a leftover from the design cut D6 rejects. That file is untouched and no such gate exists. Removed. Refs #1953 * fix(#1953): keep execute-phase.md under its byte ceiling; un-vacuum the large-file test Five of the seven remote-runner failures were one cause: the execute:post dispatch contract, written out inline, grew execute-phase.md 1876 bytes (93,400 -> 95,276) against a frozen PRE_PHASE6 ceiling of 93,600. A drift-ack does not clear that — tests/phase6-capstone-conformance.test.cjs and tests/fix-2285-claude-orchestration-wiring.test.cjs assert the file is literally under the cap. The contract now lives in gsd-core/references/loop-hook-dispatch.md, which already claimed to be the point-agnostic dispatch reference and already documented ref.skill and ref.agent. It gains the ref.command shape, its in-context validation rule, the advisory-by-construction statement, and a note that a point whose workflow hand-rolls one kind is not implementing this contract. execute-phase.md now defers to it in one line: 145 bytes of growth, 55 B of headroom under the cap. Better placement than the first cut — the reference was overstating its coverage, and this makes the claim true rather than duplicating prose next to it. Acknowledged by appending to tests/emitted-drift-acks/2930-*.json rather than a new 1953-*.json: two ack sources may never name the same path, and that fragment is already the accumulating ack for this file. Sixth and seventh failures: analyzesLargeFileWithinBounds tripped its own vacuity guard — the fixture generated ~480 KB against a `> 500000` assert, so the guard fired and the three assertions after it never ran. The test has been vacuous since it was written. The matrix row specifies ~1 MB, so N goes 8000 -> 20000 (1.17 MB, 17% margin) and the guard to > 1_000_000. Verified by reproducing the exact body against the compiled module: 1168888 bytes, 118 ms, all four assertions hold. Refs #1953 * fix(#1953): fold the execute:post step deferral into the existing resolve line The remaining two failures were one test: execute-phase.md carries a SECOND, tighter assertion than the 93,600 ceiling — `<=93400`, which is exactly its current size. The file cannot grow by a single byte. My previous fix got it under 93,600 but not under 93,400, so it still failed. ("H." in the report is just the parent describe of that same test, not a separate defect.) Rather than add a paragraph, the deferral now REPLACES the existing hook resolution line. It read: Resolve active step hooks from `EXECUTE_POST_HOOKS_JSON` where `kind == "step"` and `ref.skill == "code-review"`. which is the bug itself written down — only code-review was ever dispatched. It now reads: Dispatch each `kind == "step"` hook per @gsd-core/references/loop-hook-dispatch.md. For `code-review`: The following prose already begins "If no active code-review step hook exists", so it reads correctly and the code-review handling is untouched. Net effect on the file is -11 bytes: 93,400 -> 93,389, under the margin assertion rather than merely under the ceiling. That also removes the need for a drift-ack: the file shrank, so there is no growth to acknowledge, and the append to the shared 2930-*.json fragment is reverted. Leaving it would have shipped a claim of "145 bytes of growth" that is no longer true, on a file six other issues share. The test's own comment states the principle this ended up honoring: "the host loop must stay small — optional-feature detail belongs in the capability fragment, not the host workflow." Putting the dispatch contract in the reference rather than inline is that rule, applied. Refs #1953 * fix(#1953): keep the code-review hook literal the workflow test requires tests/code-review.test.cjs extracts the <step name="code_review_gate"> block and asserts it contains `ref.skill == "code-review"` verbatim. The previous commit replaced the line carrying that literal, so the token vanished and the test went red — a fair assertion: code-review IS the bespoke branch there and the workflow should still name it. Restored inside the same one-line deferral, which now reads: Dispatch `kind == "step"` hooks per @gsd-core/references/loop-hook-dispatch.md. `ref.skill == "code-review"`: 93,396 bytes — still under the `<=93400` margin assertion and 4 bytes below the base, so the file continues to shrink rather than grow. Because three consecutive runs were each reddened by a different assertion on this one file, this change was verified by sweeping ALL of them at once rather than one run at a time: every test under tests/ that reads execute-phase.md or references/loop-hook-dispatch.md was located by resolving its path constants, and each content/size assertion was evaluated directly against the working tree — 22 assertions, plus two real executions (gen-section-manifest --check, and emitted-attribution's full real-tree differential). All pass. That sweep also confirms the earlier judgement call: the net change to execute-phase.md is a SHRINK, and the size ratchet only gates growth, so reverting the append to the shared 2930-*.json ack fragment was correct — an ack would have been both unnecessary and factually wrong. Refs #1953 * chore(#1953): backfill changeset pr number to 3261 * docs(#1953): add the missing how-to for acting on a refactor proposal Reference and explanation shipped (COMMANDS.md, CONFIGURATION.md, FEATURES.md 159, ADR-1953) but the Diataxis how-to quadrant did not, and that is the one a user reaches for. CONTRIBUTING's required-docs table is 'new command -> COMMANDS.md + FEATURES.md', so CI was green on a gap. Enabling this feature is genuinely multi-step and no single page walked it: turn it on, tune the threshold, understand advisory vs strict, discover that strict needs a SECOND toggle on a DIFFERENT capability, and know what to do when a proposal appears. The two-toggle subtlety in particular was a footnote in a config table; here it is a section with both commands. Follows the shape of its closest siblings, resolve-edge-coverage-findings and resolve-prohibition-findings — both 'the loop surfaced a finding, here is what to do with it'. Includes a reason-code table for the silent cases, since the analyzer is deliberately quiet in six situations and a user who expected a proposal needs to tell 'nothing to report' from 'could not look'. Indexed from docs/README.md beside the other loop how-tos. Docs-only: exempt from the push gate, no re-verification, pass marker on 2af188b4 untouched. Refs #1953 * feat(#1953): warn when strict mode is on but nothing will actually block Closes acceptance criterion 5, which I had wrongly marked satisfied. refactor.trigger_strict records an untriaged proposal as an open deviation window, but a ship only STOPS if workflow.windows_enforce is also on — a toggle owned by the broken-windows capability that this feature neither sets nor requires. So a user could enable strict, believe ship was gated, and find out otherwise at ship time. The split itself stays: requires:["broken-windows"] would force-install the ledger on advisory users who never enable strict, and a ship:pre gate of our own would never fire because ship.md has no generic ship:pre gate dispatch. What was missing was discoverability, so that is what this fixes. `refactor evaluate` now emits a typed REFACTOR_STRICT_NOT_ENFORCING warning, naming the exact remediation command, whenever strict is on and either workflow.windows_enforce is off or broken-windows is unavailable. It fires only on a run that produced a candidate — with nothing to block on there is nothing to warn about, and warning every run would be noise. Reads workflow.windows_enforce through the same resolveConfigKey walk the router already uses for its own keys rather than a second config reader. Four tests cover the matrix: strict+enforce-off warns, strict+enforce-on does not, strict+ledger-absent warns, strict-off never warns. Also corrects a user-facing message in this same file that told the user to run `gsd-tools config-set` — the wrong form. docs/CONFIGURATION.md and the broken-windows capability both use `gsd config-set`, and gsd-tools is invoked as `node gsd-tools.cjs`, so the bare form may not resolve. The two adjacent messages in this file now agree. Refs #1953 --------- Co-authored-by: sim <sim@local>	2026-08-09 19:52:47 -04:00
Tom Boucher	653f95e39f	chore(#2801): remove the hostBehaviors.reviewerCli deprecated alias (#3272) ... * test(#2801): failing-first suite for the hostBehaviors.reviewerCli alias removal Inverts the Phase 5a rows that assert the derived legacy alias still contributes a reviewer slug, and adds the removal-warning coverage the alias's exit needs (ADR-2782 D9). RED against unmodified production code, by design: the six shipped manifests still declare the key and collectReviewerWarnings emits nothing for hostBehaviors. Refs #2801 * chore(#2801): remove the hostBehaviors.reviewerCli deprecated alias ADR-2782 D9, Phase 7 — the final phase of epic #2782. The derived legacy alias survived one release (Phase 5a shipped in 1.9.0; 1.9.1 and 1.10.0 have since gone out), so it goes. A declared reviewer body is now the only route onto the reviewer roster. - deriveReviewerSlugs no longer reads runtime.hostBehaviors.reviewerCli - the key is stripped from the six manifests that carried it; each already declares a reviewer body whose slug equals its capability id, so the derived roster is unchanged at the same twelve slugs - collectReviewerWarnings emits a presence-based, non-fatal removal notice for any manifest still declaring the key, reaching both the build-time registry generation and the third-party overlay load path. The check runs before the reviewer-body early-return, because the manifest it exists for is the alias-only one that has no body. - hostBehaviors stays an open, unvalidated bag for its other 59 keys; this adds one keyed removal notice, not general validation Refs #2801 * refactor(#2801): give the reviewer-warning channel a typed IR Review finding: the new tests asserted with String#includes() on the warning prose, which CONTRIBUTING.md's 'Prohibited: Raw Text Matching on Test Outputs' bans in favor of a typed intermediate representation. Adds the IR beside the renderer rather than replacing it, which is the shape that section prescribes and bin/verify-reapply-patches.cjs already models: - REVIEWER_WARNING, a frozen code enum - REMOVED_REVIEWER_CLI_FIELD, so the emitting site and its test share one symbol instead of duplicating a literal - collectReviewerWarningRecords(cap), returning typed records collectReviewerWarnings(cap) keeps its exact string[] contract as a thin map over the records, so both production consumers are untouched. Every section-K row now asserts on record.code/field/capId and none on the rendered message. Locks the code surface, asserts the renderer stays one-to-one with the records, and migrates the pre-existing Phase 2 test on the same channel off prose matching. Refs #2801 * test(#2801): invert the section F alias fall-through regression row Caught by the remote runner: 2 unique failures on both Node lanes out of 31,692. tests/reviewer-lane-declarations.test.cjs section F — Phase 5a's isolated-security-review regressions — asserted that a blank reviewer.slug falls through to the hostBehaviors.reviewerCli alias rather than dropping the lane. That is the direct inverse of this phase's contract. The original rationale held only while the alias existed. With it gone there is nothing to fall through to: a blank body is not a declaration, and a declaration is the only route onto the roster. Inverted rather than deleted — the row carries the adversarial-review provenance for the slug trim, and removing a security regression guard to make a change pass is backwards. The duplicate row added earlier in section C is dropped instead; section F is its canonical home. Also corrects two count strings Phase 5b left at eleven while asserting twelve, which would misreport on failure. Refs #2801 * docs(#2801): give the removed reviewerCli flag a migration path The Reference edit alone satisfied CI — a file under docs/ moved, so lint-docs-required.cjs was green — while the task-oriented quadrant said nothing about the removal. A maintainer whose lane had just gone silent would have found the field documented as removed and no page telling them what to do about it. Adds a migration section to the how-to: the symptom, the verbatim warning they will see, the before/after manifest, and the note to keep the reviewer slug equal to the capability id so existing review.default_reviewers entries and --<slug> flags survive. Refs #2801 * chore(#2801): backfill changeset pr number to 3272 * feat(#2801): close the runtime.hostBehaviors vocabulary ADR-1016 closes twelve descriptor axes and rejects an open escape hatch in the descriptor. It never mentioned runtime.hostBehaviors, and that silence was read as permission: 59 keys across 18 manifests, 39 of them set by a single capability, validated by nothing. The reference docs went further and attributed the open seam to ADR-1016, which does not mention the field at all. KNOWN_HOST_BEHAVIORS enumerates the vocabulary. An undeclared key yields a non-fatal UNKNOWN_HOST_BEHAVIOR record on the same D4.3 channel as the alias removal notice, reaching both build-time generation and overlay install. Warning, never error, for the reason this phase exists: an error would hard-break an out-of-tree descriptor carrying a bespoke key with no deprecation window, which is what reviewerCli was given a release to avoid. Escalation is a separate decision. reviewerCli is excluded from the unknown-key sweep so it keeps its own notice with the migration pointer rather than drawing two records. A parity test binds the vocabulary to the shipped manifests in both directions, and a second asserts no shipped capability draws a notice, so the closure is provably inert in-tree. Records the decision and the miscitation as an ADR-1016 amendment. Refs #2801 * fix(#2801): bound and sanitize the unknown-key diagnostics Two findings from an isolated adversarial review of the closure commit, both proven by execution rather than asserted. MAJOR, introduced by the closure: the new Object.keys(hostBehaviors) sweep had no ceiling. An installed third-party manifest is bounded only by MANIFEST_MAX_BYTES, and an 8.69MB manifest with 800,000 keys produced 800,000 records and ~139MB of message text, retained for the registry's lifetime in OverlayMeta.diagnostics. Now capped at ten records plus a summary carrying omittedCount, mirroring capability-loader's existing slice(0,3) idiom. The same manifest now yields 11 records and 1748 chars. MINOR, newly reachable: manifest-supplied key names were interpolated raw. Unlike cap.id, which validateCapability gates on KEBAB_RE before these diagnostics run, hostBehaviors keys have no grammar check anywhere, so ANSI escapes and CRLF reached stderr and OverlayMeta.warnings intact. New describeKey replaces C0/C1 controls and clips at 80 chars. The file already had describeValue for this and applied it only to values. Both fixes land on the pre-existing reviewer.* sweep too — it carried the identical pair, and fixing only the new copy would leave the same defect one screen from its own fix. Refs #2801 --------- Co-authored-by: sim <sim@local>	2026-08-09 19:18:56 -04:00
github-actions[bot]	4f2a4932ce	chore: sync next package version to 1.10.0	2026-08-08 05:07:32 +00:00
Tom Boucher	27aa40f65e	fix(#3023): stage pi's shared hook bundle outside pi's reserved hooks/ directory (#3175) ... * test(#3023): failing-first guard — pi must not stage hooks in its reserved dir pi reserves <configDir>/hooks as its deprecated extension location and warns on every startup when it exists. Assert a pi install stages the shared hook bundle under gsd-hooks/ instead, manifests it there, and never creates hooks/. Also adds pi to the local-scope dir table in install-shared.cjs: pi was in RUNTIME_META but not LOCAL_DIR_NAME, so scope:'local' resolved path.join(root, undefined) and no local pi install could be exercised. Fails before the fix. Verified via the remote runner. * fix(#3023): stage pi's shared hook bundle outside pi's reserved hooks/ dir pi reserves <configDir>/hooks as its now-deprecated extension location and warns on every startup when that directory merely exists — checkDeprecatedExtensionDirs() guards the warning with a bare existsSync(), unlike its tools/ sibling. GSD staged its shared hook bundle exactly there, and pi's advised remediation (move it to extensions/) would break the adapter's paths and expose GSD's .js helpers to pi's extension auto-discovery. The bundle directory name is now runtime-descriptor-driven: hostBehaviors .sharedHooksDirName, defaulting to 'hooks' so all 18 other runtimes are byte-identical. pi sets 'gsd-hooks'. The name is validated as a single path segment — separators, dot-only segments, trailing dots, absolute paths, NUL, and Windows reserved device names all fall back to the default, because the value is joined onto a user's config root and written to. Renamed in place rather than relocated: hook scripts resolve siblings via __dirname/.., so a depth change would silently break them. - install / uninstall / manifest sites all read the resolved name - pi/gsd.cjs probes gsd-hooks then hooks, so dev checkouts and half-upgraded trees still resolve; the never-throws contract is preserved - new migration 009 retires the legacy pi hooks/ dir on upgrade, using a new non-recursive remove-empty-dir engine primitive (rmdirSync only, symlink-refusing, containment-guarded); ADR-0008 amended accordingly - fixes two latent name-dependencies the rename exposed: the stale-hook scan and the injection scanner's self-exclusion both hardcoded 'hooks' Verified on the remote runner. Closes #3023 * fix(#3023): close review findings and align emitted provenance with the rename Adversarial review found two defects, and the remote runner found four failure clusters. All fixed here. Review BLOCKER — detect-custom-files was blind to the renamed bundle. GSD_PREFIX_MANAGED_DIRS in gsd-tools.cjs hardcoded 'hooks', so for pi the whole gsd-hooks/ tree was invisible to the custom-file scan and user-added files there were never backed up before the next update's clean-install wipe. The dir set now resolves via the .gsd-runtime marker plus the shipped capability registry (never bin/install.js, which is not shipped into installed trees), and falls back to scanning every known candidate when the runtime cannot be determined — over-scanning is safe, under-scanning is the data loss. Review MAJOR — the pi adapter bound to an empty bundle. resolveSharedHooksDir accepted any directory, so an interrupted install left gsd-hooks/ winning over a fully-staged legacy hooks/ and every hook silently no-opped. A candidate now qualifies only if it is non-empty. Remote-runner clusters: - emitted-provenance had no rule for the gsd-hooks/ family; added two pi-scoped rules pointing at the same sources the existing hooks/ rules use. The table is total, so an unattributed family is a hard failure by design. - pi tests in install-minimal-hooks and the install integration suite asserted the old layout; updated to derive the dir name from the descriptor rather than hardcoding either name. - 19 unrelated-looking failures on node22 only were a leaked fs mock: t.after() runs in registration order, cleanup was registered before mock.restoreAll(), and node22's JS rimraf calls the public fs.rmdirSync while node24's native path does not — so the EACCES stub leaked process-wide on one lane. Restore now runs first. Verified on the remote runner. * fix(#3023): honor PI_CODING_AGENT_DIR, ack the rename ripple, fix expandTilde pi resolves its agent dir as PI_CODING_AGENT_DIR ?? ~/<CONFIG_DIR_NAME>/agent (packages/coding-agent/src/config.ts). GSD's pi descriptor declared an empty configHome.env, so a user with that variable set had GSD installed where pi never looks. Added the env name; the dot-home-nested resolver already handled the override, so no resolver logic changed. Also fixes expandTilde in the shared runtime-homes resolver, found while adding that: it hardcoded os.homedir() and ignored the opts.home every caller threads, so EVERY runtime's tilde-valued env override (claude, antigravity, windsurf, pi) silently resolved against the real home. That is a correctness bug and a test-escape hazard — a sandboxed test asserting on a tilde override reached the developer's actual home directory. Now threaded through every branch; behavior with no injected home is unchanged. Adds the emitted-drift ack fragment for the 58 pi paths whose emitted location moved with the rename. The provenance rules satisfy the totality gate; the differential gate needs the ack because the hook sources are byte-unchanged — only the installer's target directory moved. The two hook files this branch genuinely edits stay attributed and are not double-acked. Note on piConfig.configDir: it is read from pi's OWN installed package.json (getPackageDir walks up from pi's __dirname), alongside piConfig.name — a white-label setting for a redistributed pi fork, not a per-project user setting. Documented accordingly rather than treated as an unsupported override. Verified on the remote runner. * fix(#3023): reject blank env overrides, pin adapter/descriptor parity Three review findings, all fixed. A whitespace-only config-dir override was accepted verbatim: the guard was `if (val)`, falsy only for the empty string, so PI_CODING_AGENT_DIR=' ' resolved to a literal three-space directory name instead of falling back to the descriptor default. Fixed across every env-consuming branch — dot-home, dot-home-nested, all three xdg steps, and generic-agents-root — not just pi's. Non-blank values are still never trimmed, so '~/My Agent Dir' keeps working. pi/gsd.cjs's probe list and the descriptor were two independent sources of truth for the bundle directory name; a future rename would have desynced them silently and left every pi hook quiet with no error. The probe list stays deliberate — it must resolve in a dev checkout and a half-upgraded tree, where the registry's answer would be wrong — so this adds the parity assertion the repo's generative-fix-divergence rule calls for: the descriptor value must be the FIRST candidate, and the default must remain present. Changeset body rewritten to cover the two later user-facing fixes it had not caught up with. Verified on the remote runner. * chore(#3023): backfill changeset PR number * fix(#3023): anchor injection-scan patterns and fix a macOS detection hole CI's security job flagged CONTEXT.md:124 — pre-existing prose reading 'not the same fact as a genuinely empty or absent one'. The match was the 'act as a' INSIDE 'f-act as a': the pattern had no left word boundary, so any word ending in act tripped it (fact, impact, contract, artifact, interact, redact, abstract). My four-line CONTEXT.md edit dragged the latent false positive into this PR because the scan is diff-scoped by file but reads whole files. Anchored with (^|[^[:alnum:]]) rather than rewording maintainer-owned prose, which would have left the class alive for the next PR touching any file saying 'fact as a'. Auditing the rest of the list for the same class surfaced a real detection hole: the eval/exec/Function patterns matched a quote via \x27, a GNU-grep-only hex escape. BSD/macOS grep reads it as four literal characters, so single-quoted eval('...')/exec('...') payloads were NEVER detected there while passing on GNU-grep CI. Replaced with a literal apostrophe class. Boundaries were added only where a real word-suffix collision exists; exec, jailbreak, developer mode and the role-manipulation family were audited and deliberately left unanchored. 22 new cases cover both directions — the false positives now scan clean, and every real payload still fires, including the quote/punctuation/start-of-line boundary forms. Also builds this branch's injection test fixture at runtime instead of carrying the literal phrase, so the payload keeps its teeth without tripping the scan. Verified on the remote runner. --------- Co-authored-by: sim <sim@local>	2026-08-07 13:41:21 -04:00
Dennis Kim	178ec00040	fix(#2787): clarify broken-windows ship blocking enforcement (#2814) ... * fix(#2787): clarify broken-windows ship blocking enforcement * fix(#2787): update renderLedger header to clarify opt-in enforcement * fix(#2787): address maintainer scope and wording review --------- Co-authored-by: Tom Boucher <trekkie@nomorestars.com>	2026-08-03 12:39:16 -04:00
𝚌𝚕𝚎𝚣𝚌𝚘𝚍𝚒𝚗𝚐	88f6d9bd1b	fix(#2644): deduplicate Cursor slash menu (#2812) ... * fix(#2644): deduplicate Cursor slash menu * fix: preserve installer executable mode * chore: add changeset for PR #2812 * test(#2644): acknowledge Cursor emission changes * test(#2644): drop spent emitted drift acknowledgments * fix(#2644): remove retired Cursor command converter --------- Co-authored-by: clezcoding <clezcoding@users.noreply.github.com> Co-authored-by: Tom Boucher <trekkie@nomorestars.com>	2026-08-03 12:05:45 -04:00
Tom Boucher	51f32d2d40	fix(#2658): detect trae runtime and resolve its instruction file to a concrete rules file (#3006) ... * test(#2658): add failing-first regression for trae runtime detection and instruction path Covers all three collided defects reported in #2658 plus a fourth instance of defect 1 (ingest-docs.md) found while diagnosing it: missing trae detection in workflow runtime-detection blocks, the CLAUDE.md path-mutilation bug in both the js/cjs and md install-time converters, and the missing projectInstructionFile capability declaration. Fails against current source; the next commit fixes it. * fix(#2658): detect trae runtime and resolve its instruction file to a concrete rules file Three defects collided to produce the reported ".claude/.trae/rules/" path: 1. new-project.md and ingest-docs.md's runtime-detection blocks only recognized codex/gemini/opencode and fell through to RUNTIME=claude for trae. Both now recognize the /.trae/ execution-context path and the TRAE_CONFIG_DIR env var before the claude fallback. 2. RUNTIME_CONTENT_DISPATCH.trae.js (bin/install.js) replaced bare "CLAUDE.md" before the ".claude/" prefix was handled, mutilating ".claude/CLAUDE.md" into ".claude/.trae/rules/". Now replaces the full ".claude/CLAUDE.md" path first, and targets a concrete file. 3. convertClaudeToTraeMarkdown (mirrored in bin/install.js and src/runtime-artifact-conversion.cts per the #2094 output-parity test) had the same class of bug with a different wrong output (".trae/.trae/rules/", from its generic ".claude/" rewrite firing after the bare CLAUDE.md rewrite). Both mirrors now match full-path forms before the bare/generic patterns, converging on the same concrete file as the js/cjs converter. 4. capabilities/trae/capability.json didn't declare hostBehaviors.projectInstructionFile, so getProjectInstructionFile fell through to the generic AGENTS.md default even when RUNTIME=trae was resolved correctly. Now declares ".trae/rules/rules.md", regenerated into gsd-core/bin/lib/capability-registry.cjs via npm run gen:capability-registry. Closes #2658 * chore(#2658): add changeset * fix(#2658): preserve arbitrary runtime-dir prefixes in the trae path rewrite Found by the end-to-end --trae install regression test (not by static trace) across two verification runs: 1. copyWithPathReplacement runs a generic ~/.claude/, $HOME/.claude/, and ./.claude/ -> runtime-dir rewrite on every .md file BEFORE calling convertClaudeToTraeMarkdown. The prior fix's .claude/CLAUDE.md-specific patterns never fire on that already-rewritten text, and the bare fallback still doubled whatever prefix the generic pass substituted. A first attempt handled only the fixed "./.trae/" shape and missed the $HOME/.claude/ and ~/.claude/ forms gsd-core/workflows/profile-user.md actually uses, which post-rewrite become an arbitrary absolute local-install-root path, not the fixed relative shape. Fixed with a prefix-preserving pattern that captures whatever precedes a ".trae/" tail and fixes only the filename suffix, instead of assuming one fixed shape. 2. The fix's own explanatory comments literally spelled out the malformed strings and the instruction filename as contiguous text. Since these two files ship verbatim into local --trae installs, where they are themselves run through the same find/replace, the comments got "fixed" right along with the real code, leaking the malformed string into the installed tree. Rewrote every comment in both mirror copies to never spell either the instruction filename or a malformed shape as one contiguous token. Adds an emitted-drift-ack fragment: the corrected replacement target for every CLAUDE.md mention (bare directory -> concrete file) changes trae-emitted output for every repo file that mentions CLAUDE.md, not only the ones that hit the originally reported bug. * test(#2658): extend parity test with arbitrary-prefix .trae/ inputs The bin/install.js vs runtime-artifact-conversion.cjs parity assertion for convertClaudeToTraeMarkdown only fed the pre-existing bare .claude/CLAUDE.md input through both implementations. Feed the prefix-preserving cases (relative, nested-absolute, tilde, $HOME, backtick-wrapped) plus a property-based check through both, so a future edit to only one copy of the .trae/-tail regex fails this test instead of silently diverging. * chore(#2658): backfill changeset PR number to 3006 --------- Co-authored-by: sim <sim@local>	2026-08-02 18:29:47 -04:00
kyle-the-dev	ce38d44811	fix(#2777): remove stale codex local home metadata (#2831) ... * fix(#2777): remove stale codex local home metadata * chore(#2777): add changeset for codex local layout metadata --------- Co-authored-by: Tom Boucher <trekkie@nomorestars.com>	2026-08-02 01:00:52 -04:00
github-actions[bot]	854c93533c	chore: sync next package version to 1.9.1	2026-07-31 13:12:09 +00:00
github-actions[bot]	4232a79396	chore: sync next package version to 1.9.0	2026-07-31 03:14:04 +00:00
Tom Boucher	3f6b063fbb	chore(#2799): invoke_reviewers and write_reviews iterate declared lanes (#2861) ... * chore(#2799): resolve reviewer lanes into executable invocation plans Phase 5b of ADR-2782. Adds the resolver and runner that let invoke_reviewers iterate declared lanes instead of hand-authored per-CLI bash. Five additive descriptor amendments, each forced by a lane that ships today: - LaneHandler gains 'opencode' — the lane rebuilds its review from assistant text parts of a --format json stream; a plain stdout copy re-breaks #1936. - modelConfigKey — antigravity's key is review.models.agy, not .antigravity, so resolving by slug silently dropped a configured model. - defaultHost/fallbackModel — Phase 4 federated every *_host with a default of empty string; the real fallback only existed in the bash. - args becomes an argv template with a closed four-placeholder vocabulary. Positional splicing produced 'codex --model M -o F exec --ephemeral', which is not a valid invocation: codex injects in the middle, twice. - kimi-code lane, with the bounded command-capability probe (needle --output-format) that tells Kimi Code from the legacy python kimi-cli. Parity gate re-pointed: the workflow-text families it scanned are the text this phase deletes, so they are replaced by descriptor-to-registry parity plus an anti-parity check that no bespoke leg returns. jq, curl and external timeout/gtimeout all drop out of the review path. Refs #2782 * chore(#2799): add review-lane query surface and widen the manifest vocabulary Adds the gsd-tools 'review-lane' route (plan/invoke/sections) the workflow loops over, projects all twelve lanes into their capability manifests, and widens capability-validator for the amendments. opencode admitted to VALID_LANE_HANDLERS under the second arm of the enum's own admission rule: one lane, justified by a documented upstream defect data cannot express (#1936 — the agent can end its turn with zero output tokens and --format default then drops the assistant text entirely). Two bugs caught by an end-to-end stub run and fixed here: - loadConfigResolved returns a provenance wrapper, not the config; using it directly resolved every key to undefined, which reads as 'nothing configured' and silently dropped every model override. - hasBinary used shell:true with an args array (Node 26 DEP0190). Replaced with a PATH scan that spawns nothing at all. Refs #2782 * chore(#2799): iterate declared lanes in invoke_reviewers and write_reviews Replaces the eleven hand-authored per-CLI bash legs with a loop over resolved lanes, and renders REVIEWS.md sections from each lane's declared reviewsSection instead of thirteen hardcoded headings. review.md drops from 1104 lines to 507 (61KB to 28.7KB). Parity gate re-pointed, as agreed: the leg-marker and section-heading families scanned exactly the text this phase deletes, so they are replaced by descriptor-to-registry parity in both directions, plus an anti-parity check that fires if a bespoke leg is ever re-added. Enum, emitting sites and the Object.keys lock moved together. The budget-trim helper is hoisted out of the Ollama leg: it was always lane-agnostic, and any lane may now declare a promptBudgetKey. Refs #2782 * feat(#2799): bind the consented egress host and re-verify it at invocation Completes ADR-2782 D5. Rule 1 was recorded in the ADR as delivered by Phase 3 but was not implemented: ConsentRecord had no host field and nothing in the tree bound one, so this phase's rule-4 comparison had no baseline. ConsentRecord gains an OPTIONAL reviewerHost. Optional is the whole design: isValidConsentRecord does not require it, so every record already on disk stays valid and no re-consent storm fires (D4 rule 5). It is deliberately excluded from disclosureSignature — the loader has no config resolver, so folding a config-derived value in would make loader and lifecycle compute different signatures for the same manifest and re-prompt forever. Install resolves hostConfigKey (falling back to the lane's declared defaultHost, which is what the invocation path uses) and records it. Invocation re-resolves and blocks on mismatch rather than silently redirecting. Absence allows: no record, or a record predating the field, means nothing to compare — denying there would break every existing local-model user on upgrade. Refs #2782 * test(#2799): cover the resolver, runner and handlers; retarget the parity suites Adds the golden invocation-plan table (one row per shipped lane, derived from the bash legs rather than the descriptor types) plus runner coverage for the probe, empty-output policy, the three handlers and the egress check. Retargets the existing suites onto the new contract: descriptor-to-registry parity, the anti-parity check, the opencode handler, and the twelfth lane. Two corrections found by running them: - modelConfigKey was required; that breaks D4 rule 2, since a reviewer manifest authored before this phase would fail validation on upgrade. It is optional, read as null when absent. - the antigravity non-zero-exit test pre-seeded the transcript, which asserted that a STALE entry leaks through — the exact bug the watermark prevents. The spawn now appends, as the real tool does. Refs #2782 * fix(#2799): restore agy --add-dir and the self-report prompt in the handler Retargeting the three legacy reviewer suites off the deleted bash surfaced two real regressions in the port, both #2176: - --add-dir was dropped. Without it agy's permission context never receives the cwd repo, so the agent anchors on its own scratch dir and reviews the plan text in isolation — the exact failure the Review Instructions forbid. It is capability-probed, because an older agy rejects the unknown flag outright and a lane that fails to start is worse than one running on the prompt anchor. - the prompt lost the clause mandating a REVIEWED-WITHOUT-REPO-ACCESS self-report, which is what makes a blind review distinguishable from a grounded one. antigravity now builds its own prompt variant. Also ports the #2073 mode-2 cli.log diagnostic, which was dropped: a pinned model that 404s exits 0 with empty stdout AND an empty transcript, so agy's own log is the only evidence that anything failed. The three suites now assert against the plan and the handler instead of matching fence text, so they no longer need allow-test-rule exemptions. Refs #2782 * docs(#2799): document the declared lanes, the new flag, and dropped prerequisites COMMANDS.md gains --kimi-code and replaces the jq-prerequisite paragraph, which is now false: no lane requires jq, curl or an external timeout. Adds the changed-egress-destination behavior, since a blocked lane is something a user can hit. CONFIGURATION.md records that the model config key is declared per lane rather than derived from the flag — antigravity's is review.models.agy — and adds review.models.kimi-code. reviewer-instances.md now routes an instance through its lane's single invocation seam instead of a copied per-adapter bash block, which is what lets a cross-cutting fix reach instances for free. That required implementing the --model/--agent/--as flags it documents; --model re-resolves through the lane's argv template rather than splicing, so the flag lands where the lane declares it rather than ahead of a subcommand. CONTEXT.md glossary gains both new modules. Refs #2782 * chore(#2799): drop the stale emitted-drift acknowledgment The only entry was #2797's, acknowledging COMMENT-ONLY GROWTH in review.md. That file now shrinks by ~32KB and every emitted hash that moved is attributable to this diff, so the ack no longer explains anything. Removing the last entry means removing the file: its presence is the alarm, and an empty one signals nothing. Verified by deleting it and re-running the attribution and provenance gates plus lint:ci — all green without it. Refs #2782 * docs(#2799): record the Phase 5b vocabulary widenings in ADR-2782 Five additive amendments, each forced by a lane that ships today, plus two corrections the phase had to make rather than work around: - D5 rule 1 was recorded as delivered by Phase 3 and was not implemented, so this phase's rule-4 comparison had no baseline. Recorded because an ADR asserting a rule was delivered is exactly what stops a later phase checking. - The DEFECT.GENERATIVE-FIX gate is re-pointed: its workflow-text families scanned the text this phase deletes. Also records that D7's 'skip the probe where no bounding mechanism exists' carve-out is obsolete — in practice it meant the Antigravity lane ran unbounded on every stock macOS host, which ships neither timeout nor gtimeout. Refs #2782 * fix(#2799): close four defects found by adversarial review Two confirmed bugs, both reproduced before fixing: - resolveLanePlan was not total. An openai-http lane with a missing or non-object invoke dereferenced inv.hostConfigKey and threw, contradicting the module's own documented contract; the spawn branch guarded correctly and the http branch did not. The CLI seam resolves every selected lane in one map, so one malformed overlay manifest would have aborted the whole review rather than dropping its own lane. Guarded, plus a per-lane try/catch at the seam so a throw can never take down siblings. - A reviewer-instance model was silently dropped for any lane declaring modelConfigKey null (cursor, qwen, coderabbit). reviewer_instances validates that cli is a known slug but never that the slug accepts a model, so a user could configure one, get a clean run, and never learn a different model reviewed their plan. Now warns explicitly. Two hardening fixes: - The slug is concatenated into artifact paths, so LANE_SLUG_RE is enforced in the resolver rather than inherited from a validator that does not run on this path — the module documents itself as the overlay-manifest trust boundary, so it should not depend on someone else having checked. - normalizeHost mangled a scheme-less value: new URL('localhost:11434') parses with an empty hostname, so it became 'localhost://11434' and was compared and requested as if real. An empty hostname now means not-a-URL. Also documents the one gap that cannot be closed here: the antigravity watermark is keyed by workspace, so two concurrent reviews of the same repo share a transcript. agy exposes no per-invocation id to filter on, so the handler now states which half of its never-stale guarantee actually holds. Refs #2782 * test(#2799): retarget the remaining eight review.md-asserting suites The remote runner found 37 failures the local sweep missed (it hit the shell's two-minute cap before reaching these). All eight extract per-CLI bash from review.md that this phase deletes; each protects a real invariant, so each is retargeted onto the plan, the runner or the handler rather than removed. Three real defects surfaced by doing so: - effort args never reached ANY lane. model-resolver.cjs exports no resolveExecution, so effortFor silently returned [] every time. Restored by calling the same bounded resolve-execution query the bash legs used — and NOT with --raw, which prints the resolved effort rather than the picked field, so claude got 'low' instead of '--effort low'. - the timeout guidance lost 'a silent empty output is a timeout kill, not a crash' — the operator note that exists because of the Codex 0xc0000142 misdiagnosis. Restored. - the opencode handler dropped EMPTY assistant text parts. The shipped jq was , and only substitutes for false/null — an empty string is truthy in jq and contributed a blank line. Found by a property test shrinking to ['', '']. The opencode property suite no longer spawns jq at all, which deletes the #2099 hang mechanism it was architected around rather than mitigating it. Refs #2782 * fix(#2799): register the two new generated modules, and untrack them The remote runner caught build output committed to git. Both new modules compile from src/*.cts into gsd-core/bin/lib/*.cjs, and every sibling generated that way is gitignored and eslint-ignored (ADR-457) - including Phase 1's own review-lane-descriptor.cjs. Mine were neither, so repo-invariants' "each bin/lib/*.cjs is linted xor ignored according to migration state" failed. Registered both in .gitignore and eslint.config.mjs alongside the Phase 1 module, and dropped them from the index. Nothing about the shipped behaviour changes; the artifacts are rebuilt by build:lib. This is the new-.cts-module registration ripple, and it is the one part of it I had not completed - the CONTEXT.md glossary and the inventory manifest were already done. Refs #2782 * chore(#2799): backfill changeset pr number to 2861 * chore(#2799): backfill changeset pr number to 2861 --------- Co-authored-by: Test <test@example.com>	2026-07-30 12:48:06 -04:00
Tom Boucher	0408276791	chore(#2797): federate reviewer config keys off the central schema (#2841) ... * chore(#2797): federate reviewer config keys off the central schema Phase 4 of epic #2782 (ADR-2782 D9, config half). Runs AFTER 5a per the ADR's swap amendment: a federated config slice lives inside a capabilities/<id>/capability.json, and three of the five key families had no capability directory until 5a created them. Four key families move to the lanes that use them; the central-schema removal and the federated addition land in this one commit because the exclusivity invariant fails the build on a key present in both. review.max_prompt_tokens, review.default_reviewers and review.reviewer_instances describe policy ACROSS lanes and stay central. Two things the issue did not name, both found while building it: 1. THE EXCLUSIVITY GATE WAS BLIND TO PATTERNS. It compared federated keys against manifest.validKeys only, and two of the four families (review.models.<slug>, review.max_prompt_tokens_per_reviewer.<slug>) were pattern-backed. That is not cosmetic: isCentralConfigKey consults those patterns and mergeFederatedConfig skips every key for which it returns true, so declaring a slice while the pattern survived would have shipped an INERT slice behind a green gate — the exact half-migrated shape the invariant exists to prevent. The gate now loads the patterns from the same manifest the runtime reads. 2. AN UNSET PER-LANE BUDGET NOW RESOLVES TO 0, NOT NOT-FOUND, because a federated key always resolves to its declared default. The three fallback guards in review.md checked only empty-or-"null", so a user who set the GLOBAL review.max_prompt_tokens would have silently lost trimming on the HTTP lanes. The guards now treat 0 as unset. D9 says review.models.<slug> is owned by "the lane whose slug it names". That is false for one lane: the shipped key is review.models.agy while the slug is antigravity. Ownership follows the lane; the key name is preserved, because renaming would break every config that sets it. Existing tests updated rather than left asserting the old world: config-get on a cleared federated key yields empty instead of not-found (what #2046 actually protects — never persisting the literal "null" — is unchanged and still asserted); the config-schema dynamic pattern representative moves to reviewer_instances; the prototype-pollution guard case moves to a surviving dynamic prefix so alert #26 keeps its coverage, with a new case asserting the old key is now rejected earlier; and Phase 2's harvest-widening inertness assertion becomes an ownership assertion, since Phase 4 is what consumes it. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> * fix(#2797): use a -1 sentinel so an explicit per-lane budget of 0 survives A federated config key always resolves to its declared default, so an unset per-lane prompt budget needed a value the workflow could treat as 'not configured'. The first cut used 0 — which is wrong: 0 is already a LEGITIMATE per-lane budget meaning 'do not trim this lane' (the early-return guard in prepare_trimmed_prompt_for_reviewer). Treating it as unset would have silently switched a user who deliberately disabled trimming for one lane onto the global budget. The sentinel is now -1, which is not a valid token budget, so all three states stay distinguishable: unset falls back to global, an explicit 0 disables trimming for that lane, and an explicit N is used. Locked by three CLI round-trip tests. Surfaced by the isolated security reviewer before it crashed mid-run; verified independently against the shipped trim guard rather than taken on trust. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> * fix(#2797): update central-registration assertions and stay under the review.md cap The remote runner caught both; my local sweep missed the files. 1. tests/plan-review-convergence.test.cjs asserted the three local-server host keys are in VALID_CONFIG_KEYS. They are federated to their lane capabilities now, and the exclusivity invariant forbids a key living in both places. What #2306-local actually protects is that config-set ACCEPTS them, so that is what is asserted — via isValidConfigKey, the predicate config-set itself uses, which spans central and federated. A second assertion pins federated ownership, so a silent reversion back to the central schema fails too. 2. review.md exceeded the LARGE tier hard cap (62583 > 61440). That cap is a red line, not a budget to raise. The three per-lane budget guard comments were near-identical; condensed to one terse line each. 61371 bytes, 69 to spare. Real extraction to workflows/review/modes/ is Phase 5b/6 work. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> * fix(#2797): fail closed on a broken config-schema manifest; reconcile stale docs Isolated security review findings. MAJOR — loadCentralConfigPatterns failed OPEN. It swallowed a JSON parse error and returned [], while its sibling loadCentralConfigKeys, reading the SAME file, writes to stderr and throws ExitError(1) on that identical failure class. Fail-open here defeats the gate this function exists to feed: with zero patterns, validateCrossCapability's pattern-collision check silently passes and an inert federated slice ships green. It was masked in the one production call site only because loadCentralConfigKeys runs first against the same path — a coincidence of ordering, not a guarantee, and this function is exported and called standalone. The two now share a contract: ENOENT is the legitimate absent case, anything else throws loudly. A single unparseable PATTERN is still skipped, which degrades to "checked less" rather than blocking every build. The branch had zero coverage; it now has two tests (malformed JSON, EISDIR). MINOR — docs/CONFIGURATION.md still listed review.models.qwen and review.models.cursor as settable, ~770 lines below this PR's own new Ownership section. Those lanes take no model flag, so they declare no model key and config-set now rejects them. Rows removed; the missing review.models.agy row added; the per-reviewer budget row corrected to name only the lanes that own a budget key, and to document that a per-lane 0 disables trimming for that lane. Also fixes a shadowed "raw" binding introduced by the fail-closed change, which made the generator unrequirable — caught immediately by its own --check. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> * chore(#2797): backfill changeset pr number to 2841 * fix(#2452): make the base-ref mutation test hermetic against leaked GIT_* env tests/mutation-workflow-base-ref.test.cjs fails on PR branches while next stays green, and it is currently blocking at least three unrelated PRs (#2841, #2832, #2827) with: error: invalid object 100644 <sha> for 'base-N.txt' error: Error building trees The existing loop comment attributes this to `git add .` rehashing O(n^2) blobs "before the object write had landed" and works around it by staging one path per iteration. That is not the cause: sequential execFileSync calls cannot race each other's object writes, and the failure persisted after that change — it simply moved to a lower commit index. The cause is that the git() helper inherited the runner's environment. A leaked GIT_INDEX_FILE makes `git add` write into a DIFFERENT repository's index; GIT_OBJECT_DIRECTORY / GIT_ALTERNATE_OBJECT_DIRECTORIES send the blob to another object store; GIT_DIR / GIT_WORK_TREE redirect the whole operation. In every case `git commit` then cannot resolve a blob it just staged, which is precisely the error above. Verified by negative control: with GIT_DIR exported, this test fails on the unfixed helper (the git commands operate on the wrong repository entirely); with the helper stripping GIT_* it passes. The single-path staging is kept — it is genuinely less work — but it is no longer load bearing. Found while shipping #2797. Fixed in place rather than deferred: it is a defect surfaced during the work, and it is blocking other contributors. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> * fix(#2452): build the base-advance commits empty, removing the lost-object class The base-ref guard has been failing in CI with: error: invalid object 100644 <sha> for 'base-N.txt' error: Error building trees It is currently red on at least three unrelated PRs (#2841, #2832, #2827) while next stays green. Two theories have now been tried and neither held. #1881 blamed `git add .` rehashing O(n^2) blobs and switched to staging one path per iteration; the failure moved from commit 32 to commit 25 and carried on. The preceding commit here made the git helper hermetic against leaked GIT_* environment — that IS a real vulnerability (with GIT_DIR exported the helper operates on the wrong repository entirely, proven by negative control) but it produces a different error than CI reports, so it is not demonstrably the cause either. Neither trigger reproduces off-CI, so this stops guessing at the trigger and removes the failure CLASS instead. The loop needs base-branch DEPTH and nothing else: no assertion reads these commits' contents, and base-side files cannot appear in `origin/base...HEAD` regardless. `--allow-empty` writes no blob and no tree, so there is no object for the index to reference and lose. It is also far less work than 60 write+hash+index cycles. The guard still proves its mechanism: the test asserts that a --depth=1 base fetch FAILS and a full fetch resolves, so a broken topology would surface immediately rather than passing vacuously. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5 <noreply@anthropic.com> Co-authored-by: Test <test@example.com>	2026-07-30 08:52:56 -04:00
Tom Boucher	6a9babda69	chore(#2798): declare the eleven reviewer lanes as manifest data (#2837) ... * chore(#2798): declare the eleven reviewer lanes as manifest data Phase 5a of epic #2782, delivering ADR-2782 D9 (roster half) and D3. - Five reviewers GSD never installs into become lane-only role:reviewer capabilities with no runtime body, no runtimeCompat and no install surface: gemini, coderabbit, ollama, lm-studio, llama-cpp. Before this they had no descriptor at all and lived as a hardcoded NON_RUNTIME_REVIEWER_SLUGS tail, which is now deleted outright. - The six hosts that are ALSO reviewers gain a reviewer body alongside their runtime body. Their runtime bodies are byte-identical to next -- verified per capability against the git blob, not asserted -- so no install behaviour moves. - KNOWN_REVIEWER_SLUGS derives from declared bodies via an exported deriveReviewerSlugs(registry). hostBehaviors.reviewerCli survives as a derived legacy alias for one release; where a capability carries both, the body wins and the slug appears once. Alias removal is Phase 7 (#2801). THE KEYSTONE: the roster is the SAME ELEVEN SLUGS as before -- antigravity, claude, coderabbit, codex, cursor, gemini, llama_cpp, lm_studio, ollama, opencode, qwen. This phase changes HOW the roster is derived, not WHO is in it, and the test asserts that literal list rather than a count. kimi-code is deliberately NOT declared here. It is net-new with no invoke_reviewers leg, so declaring it now would make it selectable but not invocable -- present in --all, selected, emitting an empty section for the whole 5a-to-5b window -- and would break Phase 1's parity assertion. It lands in 5b alongside the iteration that can run it. Legacy kimi (the Python CLI) is not a reviewer at all and gains nothing. The highest-value test is declaredManifestLanesMatchThePhase1Descriptor: it deep-compares all eleven declared bodies against REVIEWER_LANES field-by-field, including probe and invoke sub-fields. All eleven are byte-identical, key order included. The epic's premise is that the manifest and the core descriptor describe the same lane with NO translation layer, and Phase 2's review already caught one divergence that every other test missed. Two ADR corrections folded in, as Phases 1-3 each did: 1. PHASE ORDER. The ADR runs Phase 4 (federated config) before 5a and #2798 claims a dependency on 4. That is inverted and makes Phase 4 unsatisfiable: D9 assigns review.<host>_host to lane capabilities that do not exist until THIS phase creates them, and a federated config slice must live inside capabilities/<id>/capability.json. Real graph: Phase 2 -> 5a -> 4. 2. #2798's INVENTORY acceptance item is vacuous. The inventory catalogs bin/lib/*.cjs modules, not capability directories -- antigravity, opencode and qwen appear zero times in it -- and gen-inventory-manifest --check passes with the five new dirs and no edit. Also corrected a stale line in Phase 2's own ADR amendment: it recorded the slug pattern as /^[a-z][a-z0-9_-]*$/, but Phase 2's security review widened the shipped pattern to /^[a-z0-9][a-z0-9_-]*$/ to match Phase 1's exported LANE_SLUG_RE. The prose had not followed the code. Closes #2798 * fix(#2798): catalogue reviewer capabilities in the generated matrix The capability matrix rendered exactly two tables, feature and runtime, via renderTable(caps, role) filtering on c.role === role. ADR-2782 D3 added a THIRD role, so every role:"reviewer" capability was silently dropped from the first-party catalogue. The drift guard did not catch it, and could not: --check compares generated output against the committed file, and both omitted the five lanes identically, so it reported "up to date" while five shipped capabilities were invisible in the one document that is supposed to list what ships. A guard blind to an entire role is not guarding. This phase is what exposed it -- it ships the first role:"reviewer" capabilities -- so it is fixed here rather than deferred (CLAUDE.md: a defect found while working is fixed in the current change, which overrides one-concern-per-PR). Verified red-before-green: with a lane row deleted from the matrix, --check now exits 1; restored, it exits 0. Before this fix the lanes were absent entirely, so there was nothing for the guard to compare. Phase 6 (#2800) still owns enriching the matrix with lane-specific detail (slug/flag/transport columns) and the locale parity gate. This is the narrower fix: the capabilities APPEAR at all. * fix(#2798): close two hardening gaps and record three limits durably Isolated security review (5 targets, no blockers) reproduced two gaps in the new deriveReviewerSlugs. Both are unreachable through the checked-in registry -- it is generated, JSON-sourced and code-reviewed -- but the function is EXPORTED for reuse and carries no other validation, so it must not depend on its caller. - A whitespace-only slug passed the length>0 test verbatim and occupied a roster entry it could never match. Slugs are now trimmed before the emptiness test. A blank body correctly falls through to the legacy alias rather than DROPPING the lane, which would have been worse than the blank slug. - KNOWN_REVIEWER_SLUGS is computed at require() time, so an uncaught throw there breaks import for EVERY consumer rather than degrading selection. It is now guarded, yielding an empty roster on a malformed registry. That is a visible degradation, not a silent one: under D4 an explicitly requested reviewer that is unavailable is an ERROR, so /gsd:review --claude against an empty roster fails loudly. This also removes an asymmetry -- the sibling capability-trust module documents its collectors as TOTAL and wraps them for exactly this reason. Also records three findings that previously existed ONLY in squash-merged PR bodies, which is not a durable record: - ADR-2782 D5 gains an implementation note explaining why the resolved host is deliberately EXCLUDED from the disclosure signature. Rule 1 says consent binds the resolved host; the loader has no config resolver, so folding it in would make the loader and lifecycle compute different signatures for one manifest and re-prompt forever. The binding is split: signature covers the SHA-pinned manifest fields, the consent record stores the resolved host, and Phase 5b re-resolves at invocation -- which is where rule 4 already puts the check. A reader comparing rule 1 to the code would otherwise conclude it is unimplemented. - CONTEXT.md's capability-trust entry still described THREE executable surfaces. Phase 3 added the fourth and made that false; corrected here, since it is drift this epic introduced rather than Phase 6's new-glossary-term work. - stableJson documents the NaN/Infinity/undefined -> null signature collision and why it is unreachable (JSON grammar has no such literal, so JSON.parse throws first). Reachability rests entirely on the ingest path staying JSON.parse-only, so the note lives where someone would break it. * chore(#2798): backfill changeset pr number to 2837	2026-07-29 16:58:02 -04:00
Tom Boucher	c07734216f	fix(#2603): document kimi-code in the host-integration capability matrix (#2687) ... * fix(#2603): document kimi-code in the host-integration matrix; correct 3 inherited axes The matrix — ADR-1239's deployment source-of-truth — had a section for 18 of 19 installed runtimes but none for `kimi-code`, so its `hostIntegration` axes shipped with no citation and no evidence quote. Sourcing every axis independently against Kimi Code CLI's own docs (the issue's explicit requirement — `kimi` and `kimi-code` are distinct products) showed three values had been inherited from the Python `kimi` descriptor rather than sourced: - `embeddingMode` imperative -> declarative. Kimi Code plugins are a `kimi.plugin.json` manifest plus markdown Skills with no in-process programmatic API (docs/en/customization/plugins.md) — the same shape as `codex`. - `dispatch.nested` false -> true. The `coder` built-in "can dispatch its own nested sub-agents when a task decomposes naturally" (docs/en/customization/agents.md). The Python `kimi` CLI genuinely prohibits nesting; Kimi Code does not. - `dispatch.maxDepth` 1 -> "undocumented". Nesting is documented but no depth bound is published, so the fail-closed sentinel applies over a guessed integer. `dispatch.namedDispatch` deliberately stays `false`: GSD's kimi-code artifact layout installs Agent Skills only (no `agents` kind), so no named GSD subagent is registered with the host and `resolveDispatchType` maps every role onto coder/explore/plan. Flipping it would reintroduce the dispatch failure recorded in docs/migration/kimi-to-kimi-code.md. The matrix records the host-capability nuance under Documentation gaps instead. Behaviourally inert: `namedDispatch:false` already caps nested/maxDepth/background/ backgroundDispatch to false/0 in the effective axes (host-integration.cts:493-499), and the install adapter is not selected by `embeddingMode` (install.js:543 always uses the imperative adapter). The one visible effect is the curated profile pin, which moves programmatic-cli -> declarative-cli. Also fixes the axes legend, which omitted the `built-in-only` subagentToolkit member that has been in the closed vocabulary since kimi-code shipped. Same defect class and countermeasure as #2598: pin the corrected values and require the matrix to agree with the descriptor, because a descriptor/matrix disagreement is how the gap survived. Closes #2603 Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01QT3ibz5qJuDuGqpTGRYVGf * fix(#2603): report the maxDepth `undocumented` sentinel as a sentinel, not as malformed Surfaced by the orthogonal review of this change. `negotiateHostCapabilities` emits a sentinel-specific warning for every dispatch sub-axis carrying the documented `undocumented` value — namedDispatch, nested, background, subagentToolkit, backgroundDispatch, isolation — except `maxDepth`, which fell through to the numeric guard and reported `host dispatch.maxDepth is missing or not a number — treating as 0`. That message is indistinguishable from a genuinely malformed descriptor, so a correctly fail-closed descriptor reads as broken. Six shipped runtimes carry the sentinel here (antigravity, augment, opencode, trae, windsurf, zcode) and this PR's kimi-code correction adds a seventh, which is why it is fixed here rather than left in place. The numeric guard keeps firing for genuinely malformed values; both paths still degrade `effective.dispatch.maxDepth` closed to 0. Covered by three tests, including the boundary case that the sentinel carve-out must not swallow a real malformed value. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01QT3ibz5qJuDuGqpTGRYVGf * chore(#2603): backfill changeset PR number (#2687) --------- Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>	2026-07-26 22:52:57 -04:00
Tom Boucher	a3853472de	fix(#2598): declare OpenCode subagent dispatch synchronous, not background (#2682) ... * fix(#2598): declare OpenCode subagent dispatch synchronous, not background capabilities/opencode/capability.json advertised dispatch.background: true and dispatch.backgroundDispatch: true. negotiateHostCapabilities and every degradationFor / shouldFlattenDispatch consumer trusts these per-field values, so declaring a capability the host lacks OVERSTATES it — the opposite of the fail-closed posture the negotiation is built for. The issue's own citations needed checking before acting: the host-integration matrix (ADR-1239's designated deployment source-of-truth) documented `true` with NEWER evidence than the issue cited, and explicitly marked the issue's sst/opencode#5887 reference as a stale snapshot superseded by #2087. git log confirms #2087 deliberately flipped these from false to true, citing OpenCode v1.15.0/v1.17 as "background subagents enabled by default in all modes". Applying the issue as filed would, on that evidence, have REGRESSED a deliberate update. So the claim was verified against current upstream rather than either document. `packages/opencode/src/effect/runtime-flags.ts` on `dev` today reads: experimentalBackgroundSubagents: enabledByExperimental("OPENCODE_EXPERIMENTAL_BACKGROUND_SUBAGENTS") `enabledByExperimental` falls back to the `experimental` flag and `bool()` defaults to false — the parameter is hidden from the model unless an operator opts in by env var. Upstream #29638 is still OPEN and confirms the session loop `tasks.pop()`s one subtask at a time. #2087's "default-on in all modes" reading does not hold against current dev. The issue's CONCLUSION is therefore right even though part of its evidence was superseded: concurrent dispatch cannot be relied on, so both fields are false. The matrix rows are corrected with the verified citation rather than reverted to the old #5887 quote, so the record shows why the value is false TODAY rather than re-asserting evidence that was legitimately superseded. Neighbouring sub-fields are untouched and pinned by test: namedDispatch, subagentToolkit, and isolation:'orchestrator-worktree' (which works via `opencode run --dir` at the OS process level and is unaffected — #2584 does not depend on this value either way). Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_015TCwhbMuY37DzRMCfzTABJ * fix(#2598): re-pin the dispatch contract tests to synchronous OpenCode dispatch gsd-test on the descriptor change came back FAILED (5 unique, both node versions). The failures were not incidental — they were deliberate contract-pin tests encoding #2087's decision, one named literally "background UPGRADE": tests/host-integration-descriptors.test.cjs - EXPECTED_FLATTEN[opencode] === false (background-eligible) - the derived background-eligible set pin tests/opencode-imperative-reference.test.cjs - "descriptor declares background dispatch true/true (v1.15/v1.17 upgrade)" - "background UPGRADE changes shouldFlattenDispatch: false now" So this is a recorded decision being reversed, not drift being corrected, and it is reversed on evidence: current upstream `dev` gates the capability behind OPENCODE_EXPERIMENTAL_BACKGROUND_SUBAGENTS (default false) and upstream #29638 (OPEN) confirms the session loop still handles one subtask at a time. The issue is filed by the maintainer and explicitly directs "update golden-parity / validator fixtures as needed", which sanctions re-pinning. Behavioral consequence, verified: shouldFlattenDispatch(opencode) now returns TRUE, so GSD serializes opencode dispatch instead of trusting concurrency it cannot get. That is the correct fail-closed direction and is safe today — no shipped GSD flow drives OpenCode background waves (per the issue), and isolation:'orchestrator-worktree' is unaffected because it works at the OS process level via `opencode run --dir`, not via the native subagent. Each re-pinned test now asserts the retracted contract in the opposite direction — feeding the #2087 axes back in must still yield "would not flatten" — so a silent re-flip of either field is caught rather than merely un-asserted. lint:ci exit 0. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_015TCwhbMuY37DzRMCfzTABJ * chore(#2598): backfill changeset pr number (#2682) --------- Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>	2026-07-26 21:47:54 -04:00
Tom Boucher	0d08c32048	fix(#2590): emit Workflow scripts the Workflow tool accepts; make the backend reachable (#2681) ... * fix(#2590): emit Workflow scripts the Workflow tool accepts; make the backend reachable Every emitted script was rejected. Four invalid constructs, the first fatal on its own, so the Workflow backend could never dispatch a wave: 1. no `export const meta = {…}` first statement -> whole script rejected 2. resumeFromRunId("<id>") -> "resumeFromRunId is not defined". It is a Workflow TOOL INPUT parameter, not a script function. The run id still reaches the caller via summary.resumeRunId, to pass as that input. 3. budget(<n>) -> "budget is not a function". `budget` is a read-only object { total, spent(), remaining() } fed by the caller's token directive; a script cannot set it. Recorded as intent in a comment. 4. parallel(agent(…), agent(…)) -> "parallel() expects an array of functions". Now parallel([() => agent(…), …]) — passing agent() results directly also started every agent eagerly, before parallel() could bound concurrency. The single-plan stage had its own branch with the same parallel() defect; both branches are now one array-emitting path. Waves also emit phase() calls whose titles match meta.phases exactly, so progress groups correctly. Two secondary defects kept the script from ever being REACHED — which is why this shipped undetected: 5. NOTHING resolved the Agent SDK version. The fragment claimed there was "no scriptable way" to introspect it and told callers to omit the flag, so gate 5 returned agent_sdk_version_unknown on every automated run while `capability state` still reported active:true. True for bash, false for Node: the router now reads the installed @anthropic-ai/claude-agent-sdk version, walking node_modules up the tree and reading package.json directly — require.resolve throws ERR_PACKAGE_PATH_NOT_EXPORTED because the SDK's exports map does not expose ./package.json. Precedence: explicit flag > GSD_AGENT_SDK_VERSION > installed. Fail-closed is preserved; an unresolvable version still declines to inline. A too-old SDK now reports the truthful agent_sdk_version_below_floor instead of unknown. 6. The runtime fallback was `--runtime > GSD_RUNTIME > 'unknown'`, diverging from the canonical `GSD_RUNTIME > config.runtime > 'claude'`, so any invocation without --runtime reported runtime_not_claude on an ordinary Claude project. Now delegates to runtime-slash.resolveRuntime. The fragment's `${AGENT_SDK_VERSION:+--agent-sdk-version "$AGENT_SDK_VERSION"}` snippet is removed rather than repaired: it was also shell-dependent — zsh does not word-split unquoted parameter expansions, so it collapsed to a single argv element, argValue() never matched, and the run failed into the same agent_sdk_version_unknown, indistinguishable from genuinely unknown. Auto- resolution removes the need for the construct entirely. Verified with the issue's own repro: no flags now reaches the version gate; an SDK above the floor yields backend:"workflow" with a script that parses as a real ES module. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_015TCwhbMuY37DzRMCfzTABJ * fix(#2590): sync generated registry, repair sibling tests, reject duplicate wave ids Findings from the isolated review, all fixed. HIGH — gsd-core/bin/lib/capability-registry.cjs was stale, and `lint:ci` was already RED because of it. The registry embeds the fragment text INLINE, so the shipped/installed copy still taught the exact broken contract this PR fixes: the old `${AGENT_SDK_VERSION:+…}` bash line and the "OMIT the flag when unknown" guidance. Regenerated. (I had read `lint:ci` by grepping its output instead of checking its exit code, so I recorded a red chain as green — checking $? now.) HIGH — three existing tests asserted the OLD broken shape and would have failed CI; none was touched by the first commit: tests/fix-2285-claude-orchestration-wiring.test.cjs — matched resumeFromRunId("…") tests/claude-orchestration.test.cjs — .includes('budget(') tests/claude-orchestration-command-router.test.cjs — .includes('budget(') Each now asserts the corrected contract: the id/pool reaches the caller via summary, and neither construct is ever CALLED. Two sibling assertions had also gone vacuous — `.includes('resumeFromRunId')` still passed, but only because the new explanatory COMMENT contains that substring, not because anything is wired. Rewritten to assert the real property. MEDIUM — duplicate wave ids were never rejected. Plan-id uniqueness was checked within a wave, but nothing checked wave ids across waves. That was harmless before; it is not now, because each wave emits a `phase("Wave <id>")` call plus a matching meta.phases entry and the tool matches titles by exact string — two waves sharing an id would collapse into one progress group and misattribute the second wave's agents to the first. Rejected at validation, with tests either side of the boundary. MEDIUM — the fragment contradicted itself (its "Manifest construction" header still listed $AGENT_SDK_VERSION as orchestrator-built) and, more seriously, never told the orchestrator to pass summary.resumeRunId as the Workflow tool's resumeFromRunId INPUT. Since this PR moves resume from a broken in-script call to a tool-invocation input, an implementer following only the fragment would have silently regressed phase-resume to a no-op. Both fixed. MEDIUM — docs/how-to/enable-claude-orchestration-workflow-backend.md and docs/explanation/claude-orchestration-capability.md documented `resumeFromRunId("<id>")` and `budget(<tokens>)` as current correct output — teaching the bug as the feature. Updated to the real contract, including the required meta block and the thunk-array parallel() form. (The changeset is `Fixed`, so the docs gate exempts this; it is corrected because it is wrong, not because a gate demanded it.) LOW — the router's top-of-file comment still described the divergent `--runtime > GSD_RUNTIME > 'unknown'` chain as current, ninety lines above the fix; and inserting resolveInstalledAgentSdkVersion had orphaned resolveDetectionArgs' JSDoc above the wrong function. Both repaired. lint:ci now exits 0 (verified by exit code, not by reading output). Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_015TCwhbMuY37DzRMCfzTABJ * chore(#2590): backfill changeset pr number (#2681) --------- Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>	2026-07-26 20:49:38 -04:00
Tom Boucher	6ad30f74b6	feat(#2584): Phase 3 — scheduler consumer + isolation adapters (#2635) ... Final phase of #2584 (ADR-1239 Codex-binding amendment). execute-phase now negotiates dispatch.isolation and dispatches through the matching adapter, so a wave's independent plans run concurrently on six runtimes instead of one — with no runtime=== branch in the scheduler. harness-worktree passes the host's declared isolation flag (claude, cursor); orchestrator-worktree creates the worktree via the Phase-2 verb and spawns the executor into it with the resolved argv/cwd (codex, opencode, kimi, kimi-code); none stays sequential. Undeclared/unknown/unresolvable isolation degrades to none — never an unisolated parallel run. Fixes two shipped Phase-2 descriptors that per-host research found would fail at spawn: kimi lacked its headless flag (would launch the interactive TUI and hang the orchestrator), and kimi-code named a non-existent binary (Kimi Code installs as 'kimi'). Adds the worktree-path root confinement Phase 2 deferred here, and leading-dash guards on the resolver's prompt/cwd matching the existing git-argument guard. Closes #2627 Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>	2026-07-25 01:50:20 -04:00
Tom Boucher	4a66d62d10	feat(#2584): Phase 2 — worktree create verb + orchestrator-exec resolver (#2625) ... * feat(#2584): Phase 2 — worktree create verb + orchestrator-exec resolver Phase 2 of the negotiated executor-isolation feature (ADR-1239 Codex-binding amendment). Two building blocks for `dispatch.isolation: orchestrator-worktree` hosts, both unconsumed — no scheduler wires them yet (that is Phase 3), so no runtime behavior changes. worktree create verb (planWorktreeCreate / executeWorktreeCreatePlan / cmdWorktreeCreate in worktree-safety.cts, routed via routeWorktree in gsd-tools.cjs): validates the wave base, creates a bounded branch+worktree, records it in the run manifest reusing record-agent 4-field entry shape, returns the executor working directory. Bounded git (10s timeout, degrade-not-throw); all manifest read/parse/validate/dedupe precedes the single git side effect (no unmanifested-orphan on a bad manifest); timeout-only best-effort partial rollback (a clean collision-exit never removes a live peer worktree); fail-closed on bad base, unsafe leading-dash / .. inputs, and malformed/mis-shaped manifest. resolveOrchestratorExec (host-integration.cts): pure descriptor->argv resolver reading the new runtime.orchestratorExec descriptor field (codex/opencode/kimi/kimi-code), fail-closed on missing/invalid shape. Validator (capability-validator.cjs) + a parity guard asserting every orchestrator-worktree host declares a resolvable orchestratorExec. Adding the create route edits the installed gsd-core/bin/gsd-tools.cjs, so the golden-install-parity fixtures for all 19 runtimes are regenerated (npm run gen:golden) — the only changed hash is gsd-tools.cjs. CONTEXT.md glossary updated; capability-registry regenerated. Behavioral tests (worktree-safety + host-integration) incl. a fast-check property test and the parity sweep. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * chore: rebuild tracked state-transition.cjs to match #2400 source The tracked compiled artifact drifted from src/state-transition.cts: #2400 (commit 2bcfaa2e2) added the progress.total_plans frontmatter sync to source but the tracked bin/lib/state-transition.cjs was never rebuilt, so the fix was not shipping to consumers of the compiled artifact. The mandatory build:lib step for Phase 2 surfaced the drift; recompiling makes the already-merged, already-changelogged #2400 fix effective. Artifact-only resync (no source/test change); drift class tracked by #2591. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>	2026-07-24 21:23:06 -04:00
Tom Boucher	ec7978c0b4	feat(#2584): add dispatch.isolation sub-field, descriptors, validator + negotiation (#2604)	2026-07-24 12:51:42 -04:00
Tom Boucher	d579daa3ed	docs(#2505): Phase 6 — migration guide + built-in-only subagent-toolkit enum (#2538) ... * docs(#2512): Phase 6 — migration guide + built-in-only subagent-toolkit enum * fix #2512: update CONTRACT-PIN for built-in-only subagentToolkit value * docs(changeset): backfill PR #2538 for Phase 6 (#2512)	2026-07-22 14:30:26 -04:00
Tom Boucher	c2a305c44d	feat(#2505): Phase 2 — kimi-code Agent Skills install layout (#2520) ... * feat(#2454): PR 2 — kimi-code Agent Skills converter + install layout PR 1 registered the kimi-code EoS descriptor with empty artifactLayout (SKIP_INSTALL_CONTRACT excluded it from the end-to-end install test). PR 2 fills in the install surface: - src/runtime-artifact-conversion.cts: new convertClaudeCommandToKimiCodeSkill function. Today it delegates to convertClaudeCommandToKimiSkill (Python kimi-cli) because Kimi Code uses the same Agent Skills format + /skill: invocation per official docs. The distinct function name lets a future divergence land cleanly if Kimi Code's skill format evolves independently. - gsd-core/bin/lib/capability-validator.cjs: add to ALLOWED_SKILLS_CONVERTERS. - capabilities/kimi-code/capability.json: artifactLayout.global now declares the skills kind with converter='convertClaudeCommandToKimiCodeSkill' + home='.kimi-code' (auto-discovered at ~/.kimi-code/skills/ per Kimi Code docs: merge_all_available_skills = true default). - tests/installer-migration-install.integration.test.cjs: REMOVE the SKIP_INSTALL_CONTRACT exclusion — kimi-code now has a full install surface. - Regenerated capability-registry + capability-matrix + golden install parity + install tree fixtures for kimi-code. * fix(#2454): wire kimi-code converter into SKILLS_CONVERTER_REGISTRY + count bump - src/install-engine.cts: add convertClaudeCommandToKimiCodeSkill to SKILLS_CONVERTER_REGISTRY so the layout-driven skills install path can dispatch off the descriptor's converter string. - tests/capability-registry.test.cjs: bump VALID_CONVERTER_NAMES count 26 → 27 (added convertClaudeCommandToKimiCodeSkill). * fix(#2454): remove home override from kimi-code skills (inherit configDir) The home:'.kimi-code' override made the install plan resolve skills dest to ~/.kimi-code/skills instead of <configDir>/skills, causing the test's temp configDir to miss the install. Removing it lets skills inherit configDir like most runtimes. * fix(#2454): kimi-code install contract surface is flat-skills (no agents) Kimi Code has NO custom named subagents (per official docs: 3 built-in coder/explore/plan only). The kimi-skills-agents surface expects agents/ gsd.yaml + subagents/*.yaml which kimi-code does not produce. Changed to flat-skills which only checks for skills/gsd-* dirs. * docs(changeset): Phase 2 kimi-code install layout Added (#2509) * docs(changeset): backfill PR #2520 for Phase 2 (#2509)	2026-07-22 00:58:58 -04:00
Tom Boucher	bf8f320083	feat(#2505): Phase 1 — EoS descriptor split (kimi-code capability.json + drift-guard registration) (#2519) ... * feat(#2454): add kimi-code as an EoS capability (Node Kimi Code CLI) PR 1 of N for #2454. Establishes the EoS descriptor foundation for splitting GSD's kimi support into two distinct products per the user's directive: - kimi (existing): Moonshot's Python kimi-cli (~/.kimi, runtime: python) - kimi-code (new): Moonshot's Node Kimi Code CLI (~/.kimi-code, runtime: node, KIMI_CODE_HOME env) Per ADR-1239 EoS, runtime behavior is driven by capabilities/<id>/capability.json descriptors, not hardcoded branches in install.js. The new descriptor uses the existing primitives (dot-home configHome, skills artifactLayout, kimi-hooks-toml hooksSurface — same TOML [[hooks]] format Kimi Code reads per its docs). Critical Kimi Code constraint reflected in the descriptor: hostIntegration.dispatch.namedDispatch: false hostIntegration.dispatch.builtInSubagents: ['coder', 'explore', 'plan'] hostBehaviors.namedSubagentsSupported: false Kimi Code's official docs confirm only 3 built-in subagents with NO custom- subagent registration (the [subagent] table only has timeout_ms). The kimi-agents YAML layout (used by Python kimi-cli) is therefore NOT in kimi-code's artifactLayout. Schema adjustments: - subagentToolkit set to 'undocumented' (the existing escape hatch); the schema enum (full/read-only) lacks a 'limited'/'built-in-only' value. A follow-up PR can extend the schema enum to add 'built-in-only' as a first-class axis value reflecting Kimi Code's documented model. Registration: - capabilities/kimi-code/capability.json (new descriptor, modeled on codex) - bin/install.js: allRuntimes array + --all list + --kimi-code flag - gsd-core/bin/shared/runtime-aliases.manifest.json: kimi-code aliases (kimi-code, kimicode, kimi_code) - src/runtime-name-policy.cts: FALLBACK_ALIASES map - gsd-core/bin/lib/capability-registry.cjs: regenerated via scripts/gen-capability-registry.cjs --write Tests: - tests/multi-runtime-select.test.cjs updated for the new runtime count (18) + new --kimi-code flag test + 'All' shortcut renumbered 18 → 19. Out of scope for PR 1 (follow-up PRs in the sequence): - Install-time decision logic (kimi vs kimi-code detection / prompt) - agent-install-check semantics for kimi-code (verify Agent Skills presence) - cmdAgentSkills fallback returning subagent prompt content - Workflow template mapping (named agents → built-in coder/explore/plan) - Migration guidance for users currently on 'kimi' who are actually on Kimi Code - Schema enum extension for subagentToolkit: 'built-in-only' Refs #2454, #2095 (EoS/kimi migration epic), ADR-1239 (EoS). * fix(#2454): complete drift-guard registrations for kimi-code runtime The drift guards caught every surface that pins runtime enumeration. Each update is mechanical, driven by the guard's named failure mode: - src/runtime-name-policy.cts RUNTIME_LABELS: 'Kimi Code' label for kimi-code - src/runtime-name-policy.cts RUNTIME_FLAG_IDS: add kimi-code to the isKimiCode predicate generator - bin/install.js runtimeMap: option '11' → 'kimi-code', renumber downstream entries (11..17 → 12..18), ALL_RUNTIMES_OPTION 18 → 19 - gsd-core/bin/shared/model-catalog.json runtimeTierDefaults: kimi-code entry (null/null/null — same as kimi, no model tier defaults until configured) - docs/reference/capability-matrix.md: regenerated via scripts/gen-capability-matrix.cjs --write (kimi-code row added) - tests/global-config-home-fragment.test.cjs GOLDEN_FRAGMENT_MAP: kimi-code → '.kimi-code' - tests/fixtures/golden-install-parity/*.json: regenerated via npm run gen:golden (the runtime-aliases.manifest.json hash changed; all 17 runtime fixtures updated) The capability-registry is already regenerated from the prior commit. * test(#2454): update drift-guard tests for kimi-code runtime registration Multiple drift guards pin runtime enumeration counts and option numbering. Each update is mechanical, driven by the guard's named failure mode: - tests/runtime-flags.test.cjs: EXPECTED_FLAGS gains isKimiCode (16 → 17); 'all 16 flags' → 'all 17 flags' in test names + messages. - tests/multi-runtime-select.test.cjs: parseRuntimeInput option renumbering cascade — kilo moves 11→12, opencode 12→13, pi 13→14, qwen 14→15, trae 15→16, windsurf 16→17, zcode 17→18, All 18→19. New single-choice test for kimi-code (option 11). Prompt test updated for new numbering. - tests/host-integration-descriptors.test.cjs: EXPECTED_PROFILES gains kimi-code → 'programmatic-cli' (terminal CLI per Kimi Code docs); EXPECTED_FLATTEN gains kimi-code → false (backgroundDispatch:true per docs, same as Python kimi/opencode). - tests/global-config-home-fragment.test.cjs: table-count test renamed 13 → 14 table runtimes (kimi-code added to GOLDEN_FRAGMENT_MAP earlier). * fix(#2454): empty artifactLayout for kimi-code (PR 1 scope) The skills kind requires a converter (existing converters are per-runtime like convertClaudeCommandToKimiSkill). PR 1 of this multi-PR sequence only registers the descriptor; the actual Agent Skills converter (and a new 'convertClaudeCommandToKimiCodeSkill' function) lands in PR 2 alongside the install-time decision logic. Empty artifactLayout.global is valid and means 'nothing to install yet via the layout seam'. Also: added kimi-code to RUNTIME_META in tests/helpers/install-shared.cjs (localDir .kimi-code, globalSuffix .kimi-code), and added Kimi Code as option 11 in install.js's buildRuntimePromptText (renumbered downstream options 11..17 → 12..18, All 18 → 19). * fix(#2454): camelCase runtimeFlags for hyphenated ids (kimi-code → isKimiCode) The runtimeFlags generator previously produced 'isKimi-code' (hyphen preserved) for the new kimi-code runtime id. Property names with hyphens are awkward for consumers (flags['isKimi-code'] instead of flags.isKimiCode). The new runtimeIdToFlagName helper folds -[a-z] boundaries to uppercase, producing the conventional PascalCase flag name. The 16 prior single-word runtime ids are unaffected (the regex finds no hyphens). * fix(#2454): update remaining drift-guard tests + gen kimi-code fixtures - tests/runtime-flags.test.cjs drift guard: use proper kebab-case conversion (isKimiCode → kimi-code, not 'kimicode') so the registry comparison doesn't false-positive on hyphenated runtime ids. - tests/multi-runtime-select.test.cjs: fix kilo/opencode/pi/qwen/trae single-choice tests for the renumbered options (kilo 11→12, opencode 12→13, pi 13→14, qwen 14→15, trae 15→16). - tests/install.test.cjs: Kilo integration option 11→12, prompt test regex updated. - tests/fixtures/golden-install-parity/kimi-code.json + install-tree/ kimi-code.json: generated via UPDATE_GOLDEN=1 + UPDATE_INSTALL_TREE=1. The kimi-code install produces the standard GSD install layout (skills, contexts, references, etc.) — 436 paths, same shape as other runtimes that have no custom converter yet. * fix(#2454): add kimi-code install contract + global config home fragment - src/runtime-name-policy.cts GLOBAL_CONFIG_HOME_FRAGMENTS: add kimi-code → '.kimi-code' so getGlobalConfigHomeFragment returns the correct path instead of falling through to the default '.claude'. - tests/installer-migration-install.integration.test.cjs RUNTIME_INSTALL_CONTRACTS: kimi-code entry (same surface as kimi for PR 1; PR 2 will specialize once the Agent Skills converter lands). - tests/multi-runtime-select.test.cjs: fix space-separated-choices test for the renumbered kilo option (11 → 12). - tests/fixtures/golden-install-parity/kimi-code.json + install-tree/ kimi-code.json: regenerated after rebasing onto current next (new planner-reversibility.md from #2471 etc. now included). * test(#2454): skip kimi-code install contract until PR 2 ships install layout The end-to-end install test (tests/installer-migration-install.integration .test.cjs) asserts every allRuntimes entry installs a runtime-specific artifact surface. PR 1 of #2454 registers kimi-code in allRuntimes + the capability descriptor + flags + labels, but the install LAYOUT (Agent Skills converter + global AGENTS.md at $KIMI_CODE_HOME/AGENTS.md) lands in PR 2. The SKIP_INSTALL_CONTRACT set marks this exclusion explicit and self-removing — PR 2 removes the entry alongside adding the install surface, restoring the contract loop to full coverage. * fix(#2454): restore compact model-catalog.json format (M1 review) Per code-review M1: my prior 'fix(#2454): complete drift-guard registrations' commit used python json.dump(indent=2) which inflated the file from 165→607 lines (every nested entry got expanded) and lost the trailing newline. The semantic change was just a 3-line kimi-code entry. Restored the original hybrid format (top-level indent=2 + inner entries' one-line style) and added kimi-code in matching form. Regenerated golden install parity + install tree fixtures since the model-catalog.json hash changed. * fix(#2454): update CONTEXT.md allRuntimes glossary (17 → 18, add kimi-code) CI lint-tests job failed on the glossary drift guard (scripts/check-glossary-refs.cjs --check): ✗ CONTEXT.md's allRuntimes enum-count sentence claims 17 values but bin/install.js's allRuntimes array has 18. ✗ CONTEXT.md's allRuntimes member list has drifted from bin/install.js (missing from CONTEXT.md's list: kimi-code). Missed in the prior commits because gsd-test does not run the glossary check (it's a CI lint-tests-only check). Updating CONTEXT.md's two claims to 18 values + kimi-code in the member list. * chore(#2505): regen capability-registry + stamp kimi-code version 1.8.0 (#2511) * docs(changeset): Phase 1 kimi-code runtime Added (#2511) * test(#2511): regen kimi-code golden parity fixture after Phase 0 guard normalization lands * docs(changeset): backfill PR #2519 for Phase 1 (#2511)	2026-07-22 00:27:35 -04:00
Tom Boucher	d39ca0c004	Merge pull request #2490 from open-gsd/docs/2481-adr-1239-effort-axis ... feat(#2481): add a negotiated effortSurface axis and wire invocation-time effort	2026-07-21 20:14:27 -04:00
github-actions[bot]	8cf747724f	chore: sync next package version to 1.8.0	2026-07-22 00:06:08 +00:00
Tom Boucher	09b535ac00	feat(#2481): add a negotiated effortSurface axis and wire invocation-time effort ... ADR-1239 gains a ninth negotiated axis, effortSurface (argv | none), declaring how a host accepts reasoning effort. ADR-443 is amended in the same change because its recorded deferral is what the axis resolves: its Unblock condition offered paths (a) and (b) and stated the choice was 'a maintainer call this file records but does not make'. Path (a) is selected and satisfied here. Before this, effort reached a runtime only through install-time channels (EFFORT_RENDERING's frontmatter/api), so reviewer CLIs spawned as subprocesses silently inherited whatever effort sat in the user's own global CLI config. The review lane now resolves one universal effort through the ADR-443 cascade and renders it per host through the negotiated descriptor. Every per-host value is documentation-sourced, never inferred: - claude argv -- verified via 'claude --help' (--effort <level>) - opencode argv -- verified via 'opencode run --help' (--variant) - codex argv -- codex-rs/exec/src/cli.rs: model_reasoning_effort is NOT a CLI flag (config.toml key only), so the global -c override is the only argv route - 15 hosts undocumented -- their docs state no reasoning setting; the sentinel fails closed rather than inheriting a profile baseline No config-file vocabulary member: the only host that ever had one (Gemini CLI's thinkingConfig) was removed as a sunset runtime by 8f2ebbe9b (#1928, PR #1996), and neither Antigravity CLI nor ZCode documents a reasoning setting. Closes #2481 Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>	2026-07-21 19:10:22 -04:00