78 Commits

Author	SHA1	Message	Date
Tom Boucher	0bb7525a62	fix(#2943): rename get-library-docs -> query-docs; correct the ctx7 fallback rationale (#2963) ... * test(#2943): parity guard against the nonexistent get-library-docs tool Second context7 naming drift after #2017 (which guarded the plugin-marketplace PREFIX). #2017's guard only checks tools: frontmatter lines, not prose bodies — which is where the broken tool NAME (get-library-docs) lived. The context7 MCP server registers only resolve-library-id and query-docs; get-library-docs is a stale copy from upstream's own README. Scans the shipped prose surface (agents/, gsd-core/references|workflows/, commands/gsd/, skills/) and fails if any artifact instructs an agent to call mcp__context7__get-library-docs. Excludes tests/ (a fixture may use the name as a negative input) and CHANGELOG/RELEASE-NOTES-LEGACY (history). Fails-first: 4 offenders today (gsd-executor.md:29, research-documentation-lookup.md:5, discovery-phase.md:68 & :104). * fix(#2943): rename get-library-docs to query-docs and correct the ctx7 fallback rationale The context7 MCP server registers only resolve-library-id and query-docs (verified against upstream packages/mcp/src/index.ts); get-library-docs is a stale name copied from upstream's own README. Four shipped prose sites instructed agents to call a tool the server does not register, so every research path that loaded the canonical reference either errored, fell through to the ctx7 CLI branch, or fabricated a result. - research-documentation-lookup.md, gsd-executor.md, discovery-phase.md (x2): get-library-docs -> query-docs, params context7CompatibleLibraryId/topic -> libraryId/query (the registered contract). - Same files' ctx7 CLI fallback rationale: the cited cause (anthropics/claude-code#13898 'strips MCP tools from agents with a tools: frontmatter restriction') was wrong on two counts — #13898 is closed and was never about tools: frontmatter. Rewritten to describe the real mechanism (custom subagents cannot see project-scoped .mcp.json; they only inherit user-scoped ~/.claude/mcp.json). The fallback itself is kept. - discovery-phase.md 'mode: code/info' dropped — query-docs takes libraryId + query only; the code-vs-concepts intent is now expressed via the query text. resolve-library-id is unchanged (still registered upstream). CHANGELOG and RELEASE-NOTES-LEGACY citations are historical record, left as-is. * chore(#2943): add changeset fragment (pr:0 placeholder) * test(#2943): widen parity-guard scan surface to docs/ (isolated-review finding) The isolated adversarial review flagged that SCAN_DIRS omitted docs/, which ships docs/AGENTS.md — agent-consumed prose carrying 8 mcp__context7__* refs. No false negative today (it uses only the wildcard), but a future banned-name addition there would slip through, recreating the exact drift this guard exists to prevent. Add docs/ to the scan surface, with an EXCLUDED_FILES set for historical record (docs/RELEASE-NOTES-LEGACY.md, CHANGELOG.md) that must not be rewritten to satisfy the guard. * fix(#2943): update shifted PROSE_ALLOWLIST line + acknowledge gsd-executor.md growth The gsd-test gate caught two real consequences of the rationale rewrite in agents/gsd-executor.md (the +2-line corrected mechanism description shifted line numbers below it): 1. tests/no-bare-gsd-tools-command-position.test.cjs: the legitimate 'gsd-tools query commit' descriptive mention moved from line 791 -> 793. Update the PROSE_ALLOWLIST entry to the new line (the mention is unchanged, just relocated by my edit above it). Without this the gate reports both a stale allowlist entry (791) and a new offender (793) for the same mention. 2. tests/emitted-drift-acks/2943-context7-tool-name.json: gsd-executor.md grew 95 bytes (the accurate mechanism rationale is longer than the wrong one-line #13898 attribution it replaces). Acknowledge the growth with the reason. Both are mandated by the gate, not optional. The rename itself (get-library-docs -> query-docs) is byte-neutral-ish; only the rationale rewrite grew the file. * chore(#2943): backfill changeset PR number 2963 --------- Co-authored-by: sim <sim@local>	2026-08-01 01:34:05 -04:00
Tom Boucher	3af1941948	fix(#2772): resolve four discuss-phase text inconsistencies (dead MAX_PASSES read, gate-prompts drift, circular auto_advance, answer_validation drift) (#2886) ... * test(#2772): structural guards for the four discuss-phase text inconsistencies * fix(#2772): resolve four discuss-phase text inconsistencies 1. auto.md: remove the dead MAX_PASSES/max_discuss_passes config read (contradicted the mandated single-pass rule + wasted a shim invocation per auto run). 2. gate-prompts.md: context-handling options now match the actual check_existing flow (Update it | View it | Skip, not Overwrite|Append|Cancel); gray-area-option no longer mandates 'Let Claude decide' (contradicts discuss-phase.md's no-cop-out rule). 3. discuss-phase.md: auto_advance fallback ends the workflow instead of routing back to the already-run confirm_creation step (circular). 4. discuss-phase-assumptions.md: re-sync answer_validation to the parent canonical block (had drifted — lost the 'Other' empty-text branch). * chore(#2772): changeset fragment * fix+test(#2772): also fix the assumptions auto_advance circularity (review minor 1) + add positive test anchors (review minor 2) The sibling discuss-phase-assumptions.md had the identical auto_advance→confirm_creation circularity; fix it the same way (end the workflow). Add positive anchors to both auto_advance tests so a re-phrased regression can't slip past. File #2885 for the dead max_discuss_passes config still advertised in settings/registry/docs (review minor 3). * fix(#2772): keep discuss-phase.md under the 32000B #717 cap + ack assumptions growth The auto_advance fixes + the assumptions answer_validation re-sync grew both files past the emitted-attribution gate (and discuss-phase.md past the #717 32000B cap). Condense the auto_advance prose in both files (discuss-phase.md now net -11, under cap; auto.md already net -4650 from the MAX_PASSES shim removal). Add discuss-phase-assumptions.md to tests/emitted-drift-ack.json for its residual +220 (answer_validation re-sync + auto_advance fix). * chore(#2772): backfill changeset PR number (2886) --------- Co-authored-by: Test <test@example.com>	2026-07-30 18:14:34 -04:00
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
Rezolv	1b41083220	enhance(#2151): probe interactive-control for loading and error states (#2575) ... * feat(#2151): probe interactive-control for loading + error states The ui-consideration-probe taxonomy mapped interactive-control to only one consideration (long-text), so a control-only UI surface (e.g. a theme toggle) lifted no loading or error consideration — the verifier never asked what a control shows while its action is in flight or when it fails, and a spec omitting those states could PASS. Add 'interactive-control' to the loading and error entries' elements in UI_TAXONOMY so control-only surfaces are probed for in-flight and failure states. empty is deliberately excluded (a control is not data-bearing; empty would be Goodhart noise). No new categories, no cue-map change, no probe-core change — a widening within the closed shape-rooted 8 (ADR-550), an independently-versionable predicate- generator adapter (ADR-857). Regenerated the reference-doc coverage table and the 19 golden-install- parity fixtures (reference-doc hash). Regression test added first (RED: interactive-control yielded only long-text; GREEN: now error+loading+long-text). Closes #2151 * feat(#2151): add changeset for interactive-control loading/error coverage --------- Co-authored-by: CI Rebase Check <ci@gsd-redux> Co-authored-by: Tom Boucher <trekkie@nomorestars.com>	2026-07-28 18:31:47 -04:00
Tom Boucher	6fb832a93d	fix(#2684): bind scan.md and ship.md model dispatch to fields their workflows resolve (#2710) ... * test(#2684): failing-first guard for unbound model= dispatch placeholders Extends the #2517 omit-on-inherit bucket with a behavioral binding guard: every model="{X}" in a workflow must name a field that workflow actually binds — an init-payload key (queried for real), a shell assignment, or a declared parse field. scan.md ({resolved_model}) and ship.md ({balanced_model}) substitute names nothing emits, so the orchestrator invents the value (ADR-1411). Also pins the shipped reference that seeded the placeholder and instructs the #2517-forbidden model="inherit". RED expected on scan.md, ship.md, and references/model-profile-resolution.md. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01DPq9ovaovP2UvSVLjD4Lso * fix(#2684): bind scan.md and ship.md model dispatch to fields their workflows resolve scan.md:85 passed model="{resolved_model}" and ship.md:486 passed model="{balanced_model}" — names neither workflow's init payload emits. init.map-codebase emits mapper_model; init.phase-op emits no model field at all. With no source for the substitution the orchestrator invents a value, so model_overrides/model_policy are silently inert at both sites — the invisible partial application ADR-1411 prohibits. - scan.md: declare the init fields in a Parse JSON line and substitute {mapper_model}, matching its sibling map-codebase.md. - ship.md: ref.agent is only known at runtime, so resolve it per hook via query resolve-model and dispatch with {HOOK_AGENT_MODEL}, following the same NAME=$(...) → model="{NAME}" convention every other shell-resolved dispatch in the corpus uses (code-review.md, secure-phase.md, ui-phase.md). - Both sites now carry the #2517 rule: omit model= entirely when the resolved value is "inherit" or empty. A bare rename would have traded a dangling placeholder for model="", which 404s on non-Claude runtimes — an agent type absent from the profile table resolves to the empty string, which is exactly ship.md's case. - references/model-profile-resolution.md was the seam: it shipped the copy-pasteable {resolved_model} snippet scan.md inherited, used the stale Task( spelling, and instructed passing model="inherit" outright. Rewritten to teach the real binding convention and the omit rule. Two defects surfaced while fixing this and fixed inline rather than deferred: 1. ref.agent originates in a capability manifest, which may be third-party. Resolving it at runtime made this the first place that value reaches a shell command, so ship.md now validates its shape before interpolating and skips the hook when it fails — matching code-review.md's existing defense-in-depth pattern. Covered by a test that runs the shipped regex against real agent names and injection payloads. 2. The reference doc's omit example first placed a literal model= inside an Agent(...) comment, which tripped the #2284 fail-closed Hermes projection guard (bin/install.js:3704) and refused the install outright. Moved out of the call span; gen:golden is green across all 19 runtimes. Guard tests extend the existing #2517 bucket: every model="{X}" in a workflow must name a field that workflow binds — an init-payload key queried for real, a shell assignment, or a declared parse field. Behavior change (Hyrum's Law): the scan mapper now runs on the catalog-resolved model rather than the session model. The omit path is unchanged. Disclosed in the changeset body. Fixes #2684 Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01DPq9ovaovP2UvSVLjD4Lso * fix(#2684): validate capability-supplied ref.agent in-context, not in the shell The first cut of the ref.agent guard was placed after the injection point it was meant to close. It instructed the orchestrator to substitute the untrusted manifest value into a shell assignment and test it there: HOOK_AGENT="<the ref.agent value>" if [[ "$HOOK_AGENT" =~ ^[A-Za-z0-9][A-Za-z0-9._-]*$ ]]; then … Substitution happens before bash parses anything, so a manifest supplying `x"; touch /tmp/pwned; echo "` yields three statements and runs the middle one unconditionally — the regex fires afterwards and protects nothing. ship.md now requires the check to run in-context, the same way the workflow already reads activeHooks ("do NOT pipe it through a shell parser"), and to skip the hook outright on a mismatch. Only a value that has already matched ^[A-Za-z0-9][A-Za-z0-9._-]*$ ever reaches a command line. The guard test asserts the ordering — the in-context requirement and the absence of any raw shell assignment — not just that the pattern rejects metacharacters, since a pattern alone was exactly what gave false assurance here. Also corrects the empty-string explanation in references/model-profile- resolution.md. model_profile:"inherit" resolves to the literal "inherit", not "" (model-resolver.cts:395), and an unknown agent takes the empty-string path via resolve_model_ids:"omit" rather than by absence alone — the doc claimed all three produced "". The workflow instructions were already correct (omit on "inherit" OR empty); only the rationale in the citable reference was wrong. Both found by the isolated adversarial review pass for #2684. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01DPq9ovaovP2UvSVLjD4Lso * chore(#2684): backfill changeset PR number Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01DPq9ovaovP2UvSVLjD4Lso --------- Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>	2026-07-27 13:51:16 -04:00
Tom Boucher	89b673e40d	feat(#2631): planner emits estimate and plan-checker surfaces the over-budget flag (#2670) ... * test(#2631): failing-first planner estimate emission and over-budget surfacing * feat(#2631): emit plan estimate and surface the over-budget split recommendation * fix(#2631): extract sizing prose to references to fit planner and plan-phase caps * fix(#2631): move estimate check to plan-checker; fix template regex and caps * fix(#2631): restore ALWAYS split literal and keep gsd_run after the launcher preamble * fix(#2631): invoke estimate-check after the launcher preamble in plan-checker * fix(#2631): stop double-applying calibration; repair COMMANDS table and stale reference * chore(#2631): backfill changeset pr to 2670 * chore(#2631): backfill changeset pr to 2670	2026-07-26 14:29:02 -04:00
Tom Boucher	46ba02acde	feat(#2630): phase-estimation module, smart-zone config key, and cli verbs (#2661) ... * feat(#2630): add phase-estimation module, smart-zone config key, and cli verbs * fix(#2630): document smart_zone_tokens, refresh golden fixtures, fix null-proto property assertions * fix(#2630): align smart_zone_tokens write/read validation and harden estimation tests * chore(#2630): backfill changeset pr to 2661	2026-07-26 01:42:47 -04:00
Tom Boucher	6ee4349272	fix(#2537): extract offer_next step to references/ (~3.3KB headroom restored) (#2642) ... * fix(#2537): extract offer_next step to references/ (~3.3KB headroom restored) * chore(#2537): backfill changeset pr to 2642	2026-07-25 05:30:20 -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	77bf21b3a6	fix(#1995): widen worktree branch regex to accept agent-<id> namespace (#2548) ... * test(#1995): regression test for agent-<id> branch namespace Add failing-first tests proving that normalizeCleanupManifestEntry and planWorktreeRecordAgent reject Claude Code's current agent-<id> isolation branches (only worktree-agent-<id> is accepted). Boundary tests cover both namespaces plus rejection cases. * fix(#1995): widen worktree branch regex to accept agent-<id> namespace Claude Code's isolation="worktree" branch naming changed from worktree-agent-<id> to agent-<id>. Widen the regex in all 7 locations from ^worktree-agent-[A-Za-z0-9._/-]+$ to ^(worktree-)?agent-[A-Za-z0-9._/-]+$ so both namespaces are accepted. Introduce a shared WORKTREE_AGENT_BRANCH_RE constant in src/worktree-safety.cts to prevent future drift. Closes #1995 * fix(#1995): update workflow guards, test assertions, and baselines Widen the branch-check regex in execute-phase.md and execute-plan.md. Update all test assertions that checked for ^worktree-agent- to expect the widened ^(worktree-)?agent- pattern. Regenerate golden-install-parity fixtures, agent-size-baseline, and workflow-size-baseline. Closes #1995 * fix(#1995): update extractCwdGuardBash sanity check for widened regex The e2e test's sanity check verified the extracted bash block contained 'worktree-agent-'. After widening to '(worktree-)?agent-', update the check to match the new pattern. * fix(#1995): widen missed workflow-guard branch check + changeset + lint fixes - hooks/gsd-workflow-guard.js: widen startsWith('worktree-agent-') to /^(worktree-)?agent-/ regex — same defect class, was missed in prior commit - tests/worktree.test.cjs: fix indentation regression from prior edit - Add .changeset/1995-worktree-agent-branch-namespace.md (pr:0 placeholder) Found by orthogonal code review (Step 4). * fix(#1995): regenerate golden + size baselines for workflow-guard change * docs(#1995): backfill changeset PR number (2548)	2026-07-23 07:36:53 -04:00
Tom Boucher	f654c24a3e	feat(#2505): Phase 4 — runtime-aware subagent dispatch (Option A; resolve-dispatch-type query) (#2525) ... * feat(#2508): Phase 4 Option A — runtime-aware subagent dispatch via resolve-dispatch-type query (#2505) * fix(#2508): prose-variant preamble (avoid scanner-tripping literals) + namedDispatch===false-only mapping * fix(#2508): remove leftover old-preamble lines (keep prose variant only) * fix #2508: prose-only reference file * test #2508: regen golden install parity after workflow preamble additions * fix #2508: remove preamble from plan-phase.md (Phase 6 capstone ceiling); regen size+golden baselines * docs(changeset): backfill PR #2525 for Phase 4 (#2508)	2026-07-22 10:22:42 -04:00
Tom Boucher	c5e0371775	feat(#1951): reversibility tagging — gate one-way-door decisions (#2471) ... * test(#1951): add failing-first tests for reversibility tagging Red phase for issue #1951 (reversibility tagging: classify decisions by undo cost, gate one-way doors behind a checkpoint:decision). Tests assert, per the issue's acceptance criteria: - discuss-phase CONTEXT.md template records a **Reversibility:** field with a rationale on captured decisions, and states it is optional - gsd-planner @-references planner-reversibility.md and stays under the 49152-char agent cap (LARGE_CAP, tests/agent-size-budget.test.cjs) - a one-way rating inserts a checkpoint:decision before the dependent task; reversible inserts none; costly is flagged but never blocks - the taxonomy defaults to reversible when unsure (checkpoint-fatigue guard) and inserting a checkpoint implies autonomous: false - docs/reference/plan-md.md documents <reversibility> as optional with all three ratings - --no-reversibility-gates parses to REVERSIBILITY_GATES=false, is injected into the planner prompt, and is advertised in the command argument-hint and help full mode (argument-hint parity) - the override suppresses the gate but still persists the rating - cmdVerifyPlanStructure accepts every rating and the absent case (additive-validator guarantee, behavioral via runGsdTools) - parity: thinking-models-planning.md #4 adopts the canonical three-level taxonomy and the binary REVERSIBLE/IRREVERSIBLE vocabulary is gone - no content loss from the planner extraction made to fit under the cap Prose-contract assertions are Red until the implementation lands. The behavioral validator assertions pass immediately — regression guards proving the validator already accepts unknown optional tags. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * feat(#1951): reversibility tagging — gate one-way-door decisions Classify planning decisions by what undoing them would cost, and give a one-way door a human beat before the agent walks through it (issue #1951, The Pragmatic Programmer Topic 15 'Reversibility'; Bezos's one-way/two-way door framing). Acceptance criteria met: - discuss-phase records an optional reversibility rating with a rationale on <decisions> entries in the phase CONTEXT.md template. Unrated decisions are treated as reversible, so existing phases are unaffected. - a one-way rating makes gsd-planner insert a checkpoint:decision before the task that implements the decision, reusing the existing checkpoint mechanism -- no new checkpoint machinery. - reversible ratings trigger no checkpoint; costly ratings are flagged in the plan but never block. - the rating persists on the task as the optional <reversibility rating=> element. cmdVerifyPlanStructure accepts every rating and the absent case; the structural validator does not reject unknown optional tags. - --no-reversibility-gates (REVERSIBILITY_GATES=false) suppresses checkpoint insertion for intentionally-unattended runs while still recording ratings -- the override changes what stops the run, not what the plan remembers. Single taxonomy, not two: references/thinking-models-planning.md #4 already shipped a binary REVERSIBLE/IRREVERSIBLE classification and is loaded by both gsd-planner and gsd-plan-checker. It is rewritten onto the canonical three-level vocabulary and now points at planner-reversibility.md as the taxonomy owner, with a parity test that fails if the surfaces diverge (DEFECT.GENERATIVE-FIX-DIVERGENCE). agents/gsd-planner.md sat 47 chars under the 49152 LARGE_CAP, so the checkpoint DO/DON'T guidance was relocated verbatim into planner-antipatterns.md -- already @-referenced from the same section for the same topic, so the planner still loads it and nothing was dropped. A test guards the relocation against content loss. Files: gsd-core/references/planner-reversibility.md (NEW, canonical taxonomy + emission rules + anti-patterns), gsd-planner.md, plan-phase workflow/command/help (flag wiring + parity), plan-md.md schema, discuss-phase context template, CONTEXT.md glossary, INVENTORY + manifest, size baselines, install goldens, plugin skills regen, changeset. Closes #1951 Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * fix(#1951): address orthogonal review findings Two isolated reviewers (correctness + security), neither of which authored the change. Every finding fixed: Security — the rationale is untrusted input (ADR-1577). It originates in conversation and flows CONTEXT.md -> planner -> PLAN.md -> executor, each hop an LLM reading the previous hop's output, with no validation on the path. planner-reversibility.md and the discuss-phase template now state it is data and never instructions, and name the </reversibility> early-termination hazard explicitly -- a rationale that closes its own element injects sibling structure the executor reads as real tasks. Four tests guard it. Correctness 1 — nothing machine-enforced the feature's own promise: a task rated one-way with no preceding checkpoint:decision validated as fully clean, so a planner error silently reopened the gap this feature exists to close. cmdVerifyPlanStructure now warns on an ungated one-way rating. A warning, not an error: <reversibility> stays additive and the plan stays valid. Four tests cover ungated (warns), gated (silent), still-valid, and reversible/costly never flagged. Correctness 2 — pass-always test. The --no-reversibility-gates parse test substring-matched the whole workflow file, and plan-phase.md prose mentions both tokens in one sentence, so it passed with the bash conditional deleted: it was testing the documentation, not the parser. Now scoped to the fenced bash blocks and matched as one physical line, with a negative control confirming prose alone cannot satisfy it. Correctness 3 — costly had no itemized emission rule, only one-way did, so two agents could diverge on whether to tag costly at all. Correctness 4 — template convention break: the example ratings were bare while every sibling field uses [...] to signal substitution, inviting an LLM to copy one-way/costly forward as boilerplate. Now bracketed. Correctness 5 — latent false-green: .includes('reversible') also matches inside irreversible/irreversibility, which appear in anti-pattern prose, so a surface that dropped the real taxonomy entry would still pass. Now word-boundary matched. ADR-857 phase-6 ceiling — the first gsd-test run caught plan-phase.md 1216 bytes over its frozen 94519 ceiling (it had 49 bytes of headroom on next). The ceiling may only rise for privileged host machinery, and reversibility gating is optional-feature logic, so the wiring was slimmed to its minimum and the explanatory prose moved to the reference files the planner already loads. plan-phase.md is now 94400 bytes -- 119 under the ceiling and 70 bytes SMALLER than on next, so the host loop shrank while gaining the feature, which is what phase 6 ratchets toward. The tracer contract (tests/tracer-bullet.test.cjs) is unchanged. Lint — fixed an unnecessary non-null assertion in verify.cts and a CRLF-fragile bare \n regex in the new test (DEFECT.WINDOWS-CRLF-TEST- PORTABILITY, the #1658/#1668/#2206/#2449/#2450 class). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * test(#1951): checkpoint fixture must carry the common task elements The gated-one-way fixture built a checkpoint:decision task from the abbreviated skeleton in gsd-planner.md, which shows only the checkpoint-specific elements (<decision>/<context>/<resume-signal>). cmdVerifyPlanStructure requires <name> and <action> on EVERY task regardless of type, so the fixture failed validation for reasons that had nothing to do with reversibility: errors: ["Task missing <name> element", "Task 'unnamed' missing <action>"] Caught by gsd-test on 14d14a39 (2 failures, both this fixture). The canonical shape is in tests/verify.test.cjs:266 — a checkpoint task carries <name>/<files>/<action>/<verify> like any other. Fixture corrected to match. Verified behaviorally against the real gsd-tools CLI across all four cases: gated one-way (valid, silent), ungated one-way (valid, warns), costly (valid, silent), absent (valid, silent). Not a product defect: the validator's every-task contract is intentional and pre-existing, and docs/reference/plan-md.md scopes its required-element list to type=auto/tracer only because those are the elements a planner must author, not because checkpoints are exempt from <name>. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * chore(#1951): backfill changeset pr number to 2471 * fix(#1951): CodeQL incomplete-sanitization + prompt-injection scan collision Both CI failures were real defects in code this PR added, not false positives. CodeQL js/incomplete-sanitization (high), reversibility-tagging.test.cjs:46 — the namesRating helper built its regex with `rating.replace(/[-]/g, '\\-')`, which escapes the hyphen but not backslash, so the escape was incomplete. It was also unnecessary: `-` carries no special meaning outside a character class. Replaced with a complete metacharacter escape (backslash included). Word-boundary behavior verified unchanged across all three ratings — notably that "irreversible" prose still does not satisfy a "reversible" match, which is the false-green this helper exists to prevent. Prompt injection scan — the checkpoint fixture used the human-verification child element inside <verify>. That tag name is a fake-instruction-boundary pattern in scripts/prompt-injection-scan.sh, and the scan runs over changed files, so copying the shape from tests/verify.test.cjs (unflagged only because it is not in this diff) tripped the gate. Switched to the documented plain-prose <verify> form. The first attempt at that fix failed the same gate a second time: the comment explaining the collision quoted the offending tag literally. The comment now names it in prose instead — the scanner does not care whether a match is code or commentary, which is the whole point of the DEFECT.PROMPT-INJECTION-SCAN-COLLISION note in CLAUDE.md. Verified locally before push: scan reports 0 findings across 57 changed files, eslint clean, and both fixtures still validate as designed (gated one-way silent, ungated one-way warns, neither errors). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * test(#1951): record measured cost and halve gsd-tools spawns The Windows shard 1/3 job timeout was traced to the sharding layer, not to this PR's assertions — see #2472. Two contributing factors were this file's own, and are fixed here. 1. tests/test-timings.json had no entry for reversibility-tagging.test.cjs, so scripts/run-tests.cjs weighted it at the table's median fallback (~315ms) for LPT chunk packing. It actually measures 5595ms — an 18x under-weight. Recorded the measured value from the green gsd-test run (max across the node22/node24 lanes, per gen-test-timings.cjs's convention). Only this one entry: a full regen churns 634 entries of run-to-run drift, and the table is explicitly advisory and un-gated, so a 637-line diff does not belong in a feature PR. 2. Each verifyPlan() spawns gsd-tools, which dominates this file's cost. Spawns cut from 9 to 6 with no coverage lost: - the ungated-one-way warning and its stays-valid assertion now share one plan instead of building the same plan twice; - the reversible/costly never-flagged-as-ungated test was strictly subsumed by the additive suite, which already runs those two ratings ungated and asserts no /reversibilit/ warning at all — and the gate warning's text contains both "reversibility" and "one-way", so the broader assertion catches it. It only re-spawned gsd-tools twice to prove the same thing. Both are symptom fixes. The shard imbalance itself (19/11/10 minutes against a 20-minute cap, from a cost-blind round-robin partition that also reshuffles downstream files whenever one is inserted) is tracked in #2472. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * test(#1951): checkpoint fixture adopts the #2444 type-branched contract Surfaced by rebasing onto next, which gained #2444 (branch plan-structure validation on task type=checkpoint:*) while this PR was in review. cmdVerifyPlanStructure no longer applies one required-element set to every task. A checkpoint:decision now requires <name> + <resume-signal> + <decision> + <options>, and is exempt from the <action>/<verify>/<done>/ <files> set that auto and tracer tasks carry. The gated-one-way fixture predated that split and failed on the new requirement: errors: ["Task 'Task 0: Confirm the on-disk format' missing <options>"] Fixture rewritten to mirror the checkpoint:decision contract exactly — real <options> with two <option> children — rather than padding it with fields checkpoints no longer need. That also drops the plain-prose <verify> the earlier revision carried purely to dodge the prompt-injection scan; a checkpoint task has no <verify> requirement at all, so the workaround is moot. Verified against the real gsd-tools CLI across all four cases: gated one-way (valid, silent), ungated one-way (valid, warns), costly (valid, silent), absent (valid, silent). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>	2026-07-21 10:44:55 -04:00
Tom Boucher	455ad49ae3	feat(#2296): config-gated provider escalation on quota-exceeded (#2458) ... * test(#2296): failing-first coverage for provider escalation on quota-exceeded Covers the provider-escalation ladder layered onto EXEC.CLASSIFY: back-compat (no escalation block without --failure-class), cap boundaries at min(max_escalations, list length) at limit-1/limit/limit+1, opt-in gating, malformed/hostile provider_escalation config, the --failure-class CLI negative matrix, config-key registration, and a fast-check budget-limit property. Red until the resolver, CLI flag, and manifest key land. Refs #2296 Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * feat(#2296): config-gated provider escalation on quota-exceeded The dynamic_routing tier ladder escalates within one provider, which does not help when that provider is what ran out of quota. Add an opt-in provider ladder layered on the existing EXEC.CLASSIFY seam. - model-resolver: resolveProviderEscalation walks dynamic_routing.provider_escalation capped at min(max_escalations, list length), reporting from/to/attempted/exhausted. Invalid entries are dropped (ADR 227 shape validation). Stays a leaf module — the quota-class policy decision is the caller's, per the CONTEXT.md contract. - agent-command-router: export a frozen AGENT_FAILURE_CLASSES so the new CLI validator cannot drift from the classifier that produces the values. - resolve-execution: --failure-class flag; emits an escalation block ONLY when passed, so the existing JSON contract is byte-identical for every caller. - config-schema.manifest: register dynamic_routing.provider_escalation. - execute-phase step 7.1: auto-escalate, honor Retry-After, fail loudly naming every model tried once the ladder is spent. Refs #2296 Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * fix(#2296): extract quota recovery to a reference fragment; regen goldens The step 7.1a addition pushed gsd-core/workflows/execute-phase.md from 93390 to 95111 LF bytes, past the frozen ADR-857 Phase 6 ceiling (hard <93600, margin <=93400) asserted by tests/fix-2285-claude-orchestration-wiring.test.cjs. The base sat 10 bytes under the margin, so no inline wording would have fit. That gate's own rationale is that optional-feature detail belongs in a fragment, not the host loop. Moved BOTH the new provider-escalation branch and the pre-existing manual recovery prompt into gsd-core/references/execute-phase-quota-recovery.md, leaving step 7.1 as a one-line pointer. execute-phase.md is now 92880 bytes — 510 SMALLER than base. Also regenerates the fixtures that legitimately moved because three shipped files changed (gsd-tools.cjs, config-schema.manifest.json, execute-phase.md): golden-install-parity + install-tree for all 16 runtimes, INVENTORY.md + INVENTORY-MANIFEST.json for the new reference, and the workflow size baseline. Refs #2296 Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * test(#2351): make the C1 orphan-reaping test load-independent tests/run-with-timeout.test.cjs C1 asserted the child heartbeat file exists after a 1s group-kill window, but the child only wrote it on the first 100ms setInterval tick. Nothing synchronized the two: on a loaded container the group is SIGKILLed before that tick lands, the file never appears, and the assertion fails for a reason unrelated to reaping. Observed failing on both linux-node22 and linux-node24. The behavior actually under test is the FREEZE assertion (heartbeat stops advancing => descendant was reaped, not orphaned). That is unaffected by sampling once more at t=0. Child now writes its first heartbeat synchronously at startup before arming the interval, and the kill window widens 1s -> 3s to cover child boot under load. Both remove the timing dependency; neither weakens what the test proves. Refs #2296 Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * chore(#2296): backfill pr:2458 in .changeset/rapid-jays-bark.md * chore(#2296): regenerate fixtures after rebase onto #2402 The rebase conflicted on the generated golden-install-parity fixtures and workflow-size-baseline.json because #2402 (b6e6a22fc) regenerated the same artifacts. Conflict resolution picked a side to unblock the rebase; a true regeneration on the combined tree then produced further drift, confirming the resolved content was stale and would have dropped #2402's fixture changes. Regenerated goldens, install-tree, size baseline, and INVENTORY-MANIFEST from the merged tree. docs/INVENTORY.md keeps BOTH new reference rows. execute-phase.md is 92782 LF bytes with both #2402's and this PR's extractions applied — under the frozen ceiling (hard <93600, margin <=93400). Refs #2296 Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>	2026-07-20 14:59:30 -04:00
Tom Boucher	b6e6a22fce	fix(#2402): honor response_language across orchestrator output + UAT checkpoint renderer (#2457) ... * fix(#2402): honor response_language across orchestrator output + UAT checkpoint renderer Replays the in-flight bot branch fix/2402-response-language-orchestrator-coverage (seven commits, never pushed) onto current origin/next as a single squashed commit. The original work was substantial and correct; this commit preserves its full scope, trimmed where rebase conflicts + workflow size budgets required it. Three independent layers where response_language was being dropped are closed: Layer 1 — orchestrator-facing directives across workflows. Adds the strong "All user-facing output in this workflow MUST be presented in {response_language}; technical terms, code, paths, and subagent prompts stay in English" directive to ~40 workflows that previously either lacked it entirely (verify-work, new-project, new-milestone, quick, manager, and ~35 more) or carried only the weak subagent-prompt-only form (plan-phase, execute-phase). The directive covers narration between tool calls and banner output, not just the AskUserQuestion prompts. Layer 2 — UAT checkpoint renderer (src/uat.cts). buildCheckpoint now accepts an optional responseLanguage parameter and renders the frame strings ("CHECKPOINT: Verification Required", "Type `pass` or describe what's wrong.") in any of 9 languages (English/Spanish/French/German/Portuguese/Japanese/ Chinese/Korean/Italian) with an alias table covering ~30 input variants (en, es, español, ja, 日本語, etc.). cmdRenderCheckpoint reads config.response_language via loadConfig(cwd) and passes it through, so the byte-for-byte block verify-work.md reprints verbatim is already localized when written — preserving the anti-injection hygiene rule at verify-work.md (the model is forbidden to translate after the fact). CJK display width is computed by East Asian Width property ranges (W/F) so the right ║ border of the banner stays aligned for full-width characters. English fallback is byte-identical to the pre-fix behavior when response_language is unset or unrecognized. Layer 3 — literal English report templates in execute-phase. The top-of- workflow directive covers all template sites (templates are a structural source, not literal output). Inline render-language notes that previously sat at each template site were removed during the squash because they pushed execute-phase.md over its frozen pre-phase-6 byte ceiling (93600 — ADR-857 Phase 6 capstone). The single top directive covers the same surface with fewer bytes. Also extends src/docs.cts and src/init.cts to propagate response_language into the init JSON bundle of the additional workflows so the directive can read it. Tests added: - tests/uat.test.cjs: buildCheckpoint with unset/unrecognized language falls back to English default; recognized language swaps only the two frame strings while structural lines stay untouched; CJK display-width regression (independent recomputation of East Asian Width W/F ranges). - tests/workspace.test.cjs, tests/docs-update.test.cjs: response_language wiring through docs.cts/init.cts. References: #2402; reporter's three-layer triage + Layer-4 follow-up; the byte-for-byte anti-injection hygiene rule at verify-work.md (the reason Layer 2 must be renderer-side, not model-translated). This is a squash of the in-flight bot branch — seven commits representing the original implementation plus its subsequent fix/CJK-padding/test/ changeset/regen cycles, none of which were ever pushed or PR'd. The squash captures the final coherent state. * chore(#2402): backfill pr:2457 in .changeset/2402-response-language-orchestrator-coverage.md * chore(#2402): regen golden + size baseline after rebase against #2315 (PR #2451) Rebase conflicts were entirely in generated artifacts (golden-install-parity fixtures + workflow-size-baseline.json). After taking theirs during rebase, regenerated cleanly against the merged source tree.	2026-07-20 14:22:27 -04:00
Behruz Nassre Esfahani	d04e287fa9	fix(#2365): stop api-coverage detector false-positiving non-API phases (#2397) ... * fix(#2365): stop the api-coverage detector false-positiving non-API phases detectApiIntegration fired on any integration verb co-occurring anywhere on a line with any API noun, treated / as a word boundary (so a first-party Next.js src/app/api/... route path matched the noun "api"), and read any capitalized word before API/SDK/REST/GraphQL as a service name behind a fixed stopword denylist (so threat-model prose like "Resolver-only API" fired). Because the verify:pre seal gate is BLOCKING, a phase touching no external API could not reach UAT without fabricating a coverage matrix. The compound rule now requires the verb and noun to share one clause (sentence punctuation and table-cell walls end a clause) within a bounded word gap. Non-prose spans are excluded before matching: fenced code (already), inline code spans (new stripInlineCode in the markdown-sectionizer seam), and path-shaped tokens. The <Service> API surface rule requires proper-noun position — a clause-initial capitalized word is ordinary English and needs dependency evidence (URL / package reference) on the same line — and rejects compound modifiers ("Resolver-only", lowercase after the hyphen). A phase that integrates no external API now has a first-class, reasoned way to say so: a COVERAGE.md containing "No external API integration: <reason>" satisfies the gate (declaration + rows is contradictory and blocks). The true-positive path is pinned by regression tests: every default-vocabulary positive still fires, including the widest word-gap pairing and the surface-rule-only shape. Fixes #2365 Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * fix(#2365): tighten api-coverage detector per Codex review (round 2) Applies the Codex review findings on the initial #2365 fix: - S-1: a COVERAGE.md "no external API integration" declaration is the human override for a fallible detector, so it must PASS even when detection still fires — but the contradiction is now SURFACED in the gate output (overridden signal count + terms) instead of passing silently. - S-2: verb/noun pairing is now a term-group nearest-pair merge walk over precomputed word ordinals (computeWordStarts / minWordGap), not a match×match cross product — a hostile line repeating one pair thousands of times stays linear instead of going quadratic. - FN-4: package-shaped inline-code spans (`stripe-sdk`, `@stripe/stripe-js`) are kept as noun/dependency evidence rather than being fully masked, so a genuine dependency reference inside code ticks still corroborates. - C-1: the <Service> API surface rule now scans every candidate in every clause; a rejected first candidate no longer shadows a later genuine service. - Cross-clause binding: a verb may bind a noun in the immediately following clause only when its own clause names a service object, within a tight gap — admits "Integrate Stripe, exposing its endpoints …" without re-admitting the unrelated-clauses false-positive class. - Internal-descriptor negative evidence ("internal Payments API", "the internal endpoint") never pairs; URL/scheme matching generalized beyond http(s). All 5 acceptance criteria still hold: the three reported false positives are clean and "integrate the Stripe API" still fires. Built .cjs committed alongside the .cts. tsc + eslint (incl. no-adhoc-markdown-parsing) + lint:regression-names clean; affected suites 256/256 green. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(#2365): retune api-coverage detector fail-closed per Codex review (round 3) Codex's second-round review found the round-2 tightening had over-corrected into FAIL-OPEN false negatives — realistic external-API prose that the BLOCKING seal gate silently let through (the catastrophic class, since a missed API surface is worse than a dismissable false positive). Retuned the detector to be explicitly fail-closed: lean toward detecting, and let the one-line COVERAGE.md "no external API integration" declaration dismiss the residual false positives. Fail-open false negatives fixed (all now detect): - F1 clause-initial `<Service> API` with a plain follower ("Stripe API for payment processing") — dropped the follower-allowlist / corroboration gate on clause-initial surfaces; a service that is not a stopword, descriptor, or compound modifier is a real name from any clause position. - F2 scheme-less external host ("api.stripe.com/v1") — a dotted host with an alphabetic final label now contributes its API nouns; a first-party route path (no dotted host) still does not. - F3 vendor's first-party SDK ("Integrate Shopify's first-party SDK") — the compound path no longer filters nouns on "internal"/"first-party" (Codex: the qualifier can describe the vendor's own API, not the consuming project's). - F4 long single integration clause — removed the word-gap cap entirely: it could not separate a 21-word genuine clause from an 18-word internal one, so the clause boundary is now the whole relationship test. - F5 lowercase cross-clause service — cross-clause binding no longer requires a capitalized "service object". New false positives fixed (all now clean): - F6 a URL token that swallowed a trailing clause comma, merging two clauses — trailing clause punctuation is kept literal so the split survives. - F7 a capitalized internal component authorizing cross-clause binding — the new gate requires a dependent elaboration, not a new coordinate clause opened by a conjunction ("…, then document…"). - F8 a protocol name read as a service ("REST API", "GraphQL API") — protocol and locality descriptors are rejected in the `<Service>` position. - Finding 9: the inline-code-span scanner was O(n^2) on pathological backtick runs; rewritten to linear via a per-length run cursor (2 MB: 4.15 s -> ~6 ms), semantics preserved (148 sectionizer tests unchanged). Net simplification: the fail-closed model removed the round-2 minWordGap / groupByTerm / follower / corroboration machinery (350 insertions vs 445 deletions across the touched files). Under fail-closed, three round-2 negative tests now correctly detect (integration verb + "internal"-qualified noun, and the distant-same-clause case); none were trek-e acceptance FPs. Verified: 1491/1491 unit tests pass; tsc + eslint (incl. no-adhoc-markdown- parsing) + lint:regression-names clean; all 8 review findings reproduced as regression tests, both directions. Built .cjs committed alongside the .cts. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(#2365): resolve round-3 Codex review findings (fail-closed, round 4) Codex's round-3 adversarial review found the fail-closed retune had introduced new holes in both directions. Resolved: Fail-open false negatives (now detect): - External host addressing a PATH ("graph.microsoft.com/v1.0/me") is itself an integration surface and contributes an endpoint noun even when the host names no vocabulary word. A bare domain link with no path ("https://example.com") stays a non-signal, so "Integrate … from example.com, document …" is still clean. - Locality qualification ("internal", "private") no longer leaks across a sentence or clause boundary: only plain spaces may separate the descriptor from the service, so "The cache is private. Stripe API …" now detects. - Cross-clause binding: the fragile head-word cap (which could not tell a genuine "Connect … to Stripe payments, exposing its endpoints" from an unrelated "Integrate … from URL, document …" — both 4 words after the verb) is replaced by a participial-continuation rule: a verb binds a noun in the next clause only when that clause begins with an "-ing" elaboration. This fixes the 4-word-head false negative AND the false positive below at once. False positives (now clean): - Cross-clause no longer binds a finite continuation regardless of separator: "Wire the settings form. Document endpoint props." / "…; document …" / "…, document …" are separate actions, not elaborations. Perf (quadratic → linear): - The trailing-punctuation peel is a backward char scan instead of an unanchored `[…]+$` regex (16k chars: 156 ms → ~1 ms). - SERVICE_SURFACE_API_RE bounds the service-name length {1,40} so a hostile "A-A-…-x" run cannot drive O(n^2) backtracking (16k: 385 ms → ~3 ms). Consumer fail-open (blocking gate): - readPhaseScope now distinguishes "no plans" from a plan that EXISTS but is unreadable. On a read error the gate BLOCKS ("could not read the phase scope …") instead of silently certifying no-integration from partial scope — an unreadable plan could be the one describing the integration. Documented fail-closed tradeoffs, now pinned with tests so they are not "fixed" back into a fail-open: a clause-initial capitalized common word before "API" ("Payment API", "Search API") reads as a service name; a long clause pairs a verb with a distant noun; and a CommonMark inline code span that wraps a newline is matched within-line only. Codex judged these acceptable because the COVERAGE.md declaration is a cheap override. One documented limitation remains out of scope: "Integrate Stripe, and authenticate requests with its API" (a coordinate finite clause whose noun refers back by pronoun) needs coreference resolution, beyond a lexical detector. Verified: 379/379 affected + command-router tests pass (+14 new regression tests covering every round-3 finding, both directions); tsc + eslint (no-adhoc-markdown-parsing) + lint:regression-names clean. Built .cjs committed. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(#2365): simplify to robust core — remove whack-a-mole heuristics (round 5) Round-4 review confirmed the detector's two most complex features generate findings in both directions no matter how they are tuned, because they need a vendor dictionary + coreference the issue rules out in principle. Per the operator's "ship the robust core" decision, both are removed and their gaps are documented rather than chased further: - Cross-clause binding DELETED (allowsCrossClause / participle rule). It caused a fail-open on finite continuations ("Integrate Stripe; use its OAuth endpoints" — missed) and a false positive on "-ing"-SPELLED nouns ("…, billing endpoint terminology…" — wrongly fired). Detection is now same-clause only. - URL-path-as-evidence REVERTED. Treating every path-bearing URL as an endpoint fired on ordinary asset/link URLs ("…/theme.css", "…?next=/x", a docs/repo link) and recreated routine UI-phase false positives. An external URL is evidence only when it NAMES an API vocabulary word ("api.stripe.com/v1"). Two fail-open cases are now DOCUMENTED limitations, pinned by tests so a future maintainer does not re-add the heuristics that caused the false positives above: a service named only in a clause separate from its API noun, and a bare external host that names no vocabulary word. Both are cheaply covered by the COVERAGE.md declaration and rare in real phase prose ("integrate the X API"). Also fixed from the round-4 review: - Qualification now survives markdown emphasis ("The **internal** Payments API" stays clean) while still not crossing a sentence/clause boundary. - readPhaseScope fail-closes on a REAL read failure (EACCES/EIO) enumerating the phase directory or reading the roadmap fallback — not only per-plan-file failures; a missing directory/section remains a legitimate no-op. The declaration-override path surfaces scope_read_error so an incomplete-scope override stays visible. - SERVICE_SURFACE_API_RE length-bound comment no longer overclaims. Net: the detector is same-clause verb+noun + `<Service> API` surface, with path/code/inline masking and a fail-closed posture. All five acceptance criteria hold. 1573/1573 unit tests pass; tsc + eslint (no-adhoc-markdown-parsing) + lint:regression-names clean. Built .cjs committed. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(#2365): close roadmap-fallback fail-open + stale JSDoc (round-5 review) The round-5 sanity review confirmed the detector simplification is sound (all acceptance positives fire, all required negatives clean) and flagged one real blocker plus a nit: - Blocker: readPhaseScope's roadmap fallback could still silently pass an UNREADABLE roadmap. getRoadmapPhaseWithFallback gated on fs.existsSync(), which returns false on EACCES/EIO too — so an unreadable ROADMAP.md read as "absent", no exception reached isRealReadFailure, and the blocking gate certified empty scope. Fixed at the source: read the roadmap directly and honor the function's OWN documented contract — null only on ENOENT (genuinely absent), otherwise throw. Both existing callers already wrap it in try/catch expecting that throw, and readPhaseScope now fail-closes (blocks) via its roadmap catch. Verified by a new e2e test (unreadable roadmap fallback → block). - Nit: the detectApiIntegration JSDoc still described the removed cross-clause participial binding and "every external hostname counts" — corrected to the actual same-clause-only behavior and the names-a-vocab-word URL rule. Verified: full unit suite green; tsc + eslint + lint:regression-names clean. Built .cjs committed (roadmap.cjs is gitignored/rebuilt, per repo convention). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * chore(#2365): backfill changeset PR number (#2397) Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * chore(#2365): sync generated capability-registry + recapture install goldens CI surfaced two generated-artifact staleness issues (all failing test shards + lint-tests traced to these, not to a logic defect): - gsd-core/bin/lib/capability-registry.cjs was stale: the initial fix edited the ai-integration `api-coverage-plan-pre.md` fragment (added the "No external API integration" declaration section) but did not regenerate the registry, which embeds an inline copy of that fragment. Regenerated via `gen-capability-registry.cjs --write` — the diff is exactly the fragment text sync. Fixes `lint:generated-sync` and the "committed registry is in sync" + "registry integration" tests. - The 18 golden-install-parity fixtures were stale by exactly one hash line each — `gsd-core/references/api-coverage.md`, which this PR edits and which is a hashed installed artifact. Recaptured with `UPDATE_GOLDEN=1`; the diff is that single hash per runtime and nothing else. Fixes the `golden parity — *` tests. No source or behavior change — generated artifacts only. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * test(#2365): flip representative-corpus manifest to assert the fixed behavior The #2371 representative corpus (merged into next after this branch was cut) is a known-bug tripwire: it asserts each fixture's currentBuggyOutput so the test fails loudly the moment #2365 is fixed, at which point — per its own contract in representative-corpus.test.cjs — the fixer removes currentBuggyOutput so the assertion checks expectedDetected instead. This is that moment. Removed currentBuggyOutput from the three detector fixtures (nextjs-route-path, unrelated-verb-noun, threat-model-prose); the corpus now asserts detected:false, which the fail-closed same-clause detector satisfies. Notes updated to describe the fix rather than the bug. The #2366 matrix corpus is left untouched — that tripwire belongs to its own PR (#2374). Corpus test: 7/7 pass. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * test(#2365): skip chmod-000 fail-closed e2e tests on Windows The three fail-closed gate tests induce an unreadable plan / directory / roadmap with chmod 000, but Windows does not enforce POSIX mode bits — readFileSync still succeeds, so the gate never reaches the read-error path and the assertion fails on the windows-latest CI leg. The fail-closed LOGIC is platform- independent (readError → block) and is fully exercised on the macOS/Linux legs; only the method of inducing EACCES is POSIX-specific. Guard the three tests to skip on win32 as well as root, mirroring golden-install-parity's win32 skip. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * test(#2365): address trek-e review — glossary, clock-seam, IO injection, bounds Review response to PR #2397 (trek-e, CHANGES_REQUESTED). Fix logic unchanged; this closes the test/process-hygiene findings. Major: - CONTEXT.md "Markdown Sectionizer" glossary now lists the two exports this fix relies on, `stripInlineCode` and `scanInlineCodeSpans` (glossary is a PR gate). - Replaced the banned wall-clock assertion in the "hostile repeated-term line" test (Clock Seams rule — no elapsed-time asserts) with a deterministic signal-count assertion, which also directly verifies the term-dedup that keeps pairing linear (one signal for a 10k-pair line, not thousands). - Rewrote the three fail-closed read-failure tests: instead of chmod 0o000 (a no-op under root / on Windows, the pattern the repo's IO-failure convention avoids) they now exercise the newly-exported `readPhaseScope` in-process and inject the failure by monkeypatching fs.readFileSync/readdirSync to throw, restoring in finally. Deterministic and platform-independent (no skip needed), and they add the ENOENT-is-absence case that the chmod tests couldn't express. Minor: - Added limit / limit+1 boundary tests for SERVICE_SURFACE_API_RE's {1,40} service-name bound, QUALIFIER_LOOKBACK's 24-char window, and REASON_MAX_LEN (200) on the declaration reason. - Added a fast-check property that fuzzes the tokenizer / clause splitter / masking (scanLineTokens, splitClauses, collectTermMatches) with adversarial tokens (slashes, backticks, URLs, clause punctuation) and asserts the detector is total (never throws), shape-stable, holds detected <=> signals, and is deterministic. readPhaseScope is exported for the in-process tests. Verified: 125 detector + 19 gate tests pass; tsc + eslint + generated-sync (glossary/registry) + lint-regression-test-names + lint-test-file-count clean. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Fable 5 <noreply@anthropic.com>	2026-07-19 14:59:25 -04:00
Tom Boucher	d0bacc2517	fix(#2351): replace hardcoded timeout with portable run-with-timeout (#2426) ... * fix(#2351): replace hardcoded gnu timeout with portable run-with-timeout Stock macOS ships neither `timeout` nor `gtimeout` (GNU coreutils). The 10 hardcoded `timeout <n> <cmd>` calls across the workflow/agent/reference gates exited 127 ("command not found") on such hosts, and the gates — which only distinguish 0/124/other — misreported a passing build or test as a FAILURE. Fix: a single Node-based `gsd_run run-with-timeout <secs> [--] <cmd…>` verb in gsd-tools.cjs. Coreutils-independent (stock macOS AND Windows), keeps GNU `timeout`'s exit-code contract (124 timeout, passthrough, 127/126 ENOENT/EACCES, 128+signum on signal), inherits stdio so pipes/redirects work, and reaps the whole process group so a watch-mode runner cannot outlive its budget. Runs before gsd-tools' flag parsing so the wrapped argv stays opaque. Hardened per adversarial review: - On timeout, SIGKILL the group SYNCHRONOUSLY before resolving — a descendant that traps SIGTERM was otherwise orphaned holding stdout, hanging captured gates (the exact watch-mode hang the feature prevents). - Forward SIGINT/SIGTERM to the child tree instead of dying and orphaning it. - Reject blank/whitespace <seconds> (was a silent unbounded run); clamp the timer to the 32-bit setTimeout ceiling (was a spurious immediate timeout). - Lint detector: catch GNU long options / `-k5` / `$((...))`; anchor to command position so prose "timeout 30 seconds" no longer false-positives. Resolution lives once in the CLI; all 10 sites call the shared verb. A parity guard (scripts/lint-portable-timeout.cjs, wired into lint:ci) fails the build if a bare `timeout`/`gtimeout` execution reappears (the portable `command -v timeout` probe form is intentionally allowed). Also fixes the identical bug in the zh-CN checkpoints translation, updates the tests that asserted the old strings, trims a redundant phrase in gsd-verifier.md to keep it under its size hard cap, and refreshes the size baselines + golden install-parity fixtures. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * chore(#2351): add changeset (#2426) * chore: regenerate golden/size baseline after rebase onto next --------- Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>	2026-07-19 14:37:15 -04:00
Tom Boucher	40ce95f882	fix(#2358): scope review.md and ship.md temp files to a per-run mktemp directory (#2433) ... * fix(#2358): scope review workflow temp files to a per-run mktemp dir /gsd-review wrote every prompt/section/output temp file to a hardcoded /tmp path keyed only on the phase number, so two GSD projects sharing a small phase number collide on the exact same path and a crashed run's leftover file becomes bait a later, unrelated run can silently read. ship.md's external peer-review stderr capture was strictly worse — one shared, unqualified path across every project/phase/run. Thread a single mktemp -d "${TMPDIR:-/tmp}/gsd-review.XXXXXX" run directory through every review.md temp path (67 sites) via a new {run_dir}/$RUN_DIR placeholder, mirroring the existing {phase} substitution mechanism, and clean it up at the end of the run. Route ship.md's stderr capture through a per-run mktemp file the same way. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * chore(#2358): regenerate fixtures + lint gate-prep * fix(#2358): repair failing tests after gate verification * fix(#2358): thread RUN_DIR scoping into reviewer-instances.md (#1517) review.md's own invoke_reviewers step lazily loads gsd-core/references/reviewer-instances.md for the review.reviewer_instances codepath, but that doc was missed when review.md and ship.md were moved to the run-scoped {run_dir} temp directory. It still read the combined prompt from the old /tmp/gsd-review-prompt-{phase}.md (which build_prompt no longer writes, breaking reviewer-instances functionality outright) and wrote each instance's output to the old unscoped /tmp/gsd-review-${INSTANCE_NAME}-{phase}.md, leaving the exact cross-project temp-file collision bug open for that code path. Both paths now thread through {run_dir}, matching every other reviewer block in review.md. Extends the existing #2358 regression test with assertions pinning reviewer-instances.md's prompt read and output write to {run_dir}, and adds the Fixed changeset fragment. Regenerated the golden-install-parity content hashes for reviewer-instances.md via `npm run gen:golden` (paths unchanged; only the modified file's hash moved). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * chore(#2358): backfill changeset pr (#2433) --------- Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>	2026-07-19 13:12:03 -04:00
Tom Boucher	1720aacf0c	feat(#1949): <precondition> task element — Design by Contract (#2422) ... * test(#1949): add failing-first tests for <precondition> element Red phase for issue #1949 (Design by Contract: <precondition> element asserted before task execution). Tests assert: - docs/reference/plan-md.md documents the new <precondition> element - agents/gsd-planner.md @-references planner-preconditions.md and stays under the 49152-char cap (progressive-disclosure requirement) - gsd-core/references/planner-preconditions.md exists and documents the three emission cases mandated by the issue (user_setup / prior-phase artifact / env-var) and the contract triad mapping - agents/gsd-executor.md asserts <precondition> before task execution and routes unmet preconditions through existing checkpoint machinery - cmdVerifyPlanStructure (behavioral via runGsdTools) accepts plans both with and without <precondition> — the additive-validation guarantee - Parity assertion: plan-md.md and planner-preconditions.md agree on the canonical tag spelling (DEFECT.GENERATIVE-FIX-DIVERGENCE guard) Most prose-contract assertions are Red until the implementation lands. The behavioral validator assertions pass immediately (regression guards proving the validator already accepts unknown optional tags). * feat(#1949): <precondition> task element — Design by Contract Add an optional <precondition> element to <task> in PLAN.md (issue #1949, The Pragmatic Programmer Topic 23). The front-of-task side of the plan contract — preconditions (before) ↔ postconditions (<verify>/<done>/ <acceptance_criteria>, after) ↔ invariants (must_haves.truths, across the whole plan). Together with the tracer-bullet proposal (#1945), this closes both ends of the 'outrunning your headlights' failure mode for an autonomous AI executor. Acceptance criteria met: - <precondition> is an optional element on <task>; plans that omit it validate unchanged (cmdVerifyPlanStructure checks for presence of required tags, does not reject unknown optional tags). - gsd-executor evaluates the precondition before any other task work. Unmet halts execution with a checkpoint:human-verify and no partial commit; met or absent produces no visible change to execution flow. Unmet is never auto-approved under AUTO_CFG=true — a missing prerequisite is a fact the executor cannot establish on its own. - gsd-planner emits <precondition> in exactly the three cases the issue mandates: user_setup consumption, prior-phase artifact dependency, and env-var/runtime-config dependency. - Tests cover met, unmet, and absent preconditions plus the additive- validator guarantee. Files: - gsd-core/references/planner-preconditions.md (NEW): full emission rules, the three cases with worked examples, format guidance, anti-patterns, the contract triad mapping, and the executor assertion contract. Progressive disclosure. - agents/gsd-planner.md: slim <precondition> note in Task Anatomy with @-reference to the new file. To stay under the 49152-char agent-file cap (27-char headroom before this change), the inline <comment_text_discipline> and <region_scoped_negative_gate> summaries are compressed to one-line pointers — their full rules already live in planner-antipatterns.md, so no content is lost. - agents/gsd-executor.md: new step 0 'Precondition check' in the execute_tasks loop, before the type dispatch, routing unmet through checkpoint_return_format. - docs/reference/plan-md.md: new Preconditions section in the schema reference, with the canonical example and the three emission cases. - CONTEXT.md: Precondition glossary entry as a sibling of Tracer Bullet. - docs/INVENTORY.md + INVENTORY-MANIFEST.json: row for the new references/planner-preconditions.md (regen via gen-inventory-manifest). - tests/precondition-element.test.cjs: failing-first tests covering schema docs, planner emission contract, executor assertion contract, reference-file presence + the three cases, behavioral additive- validator guarantee, and a parity assertion (DEFECT.GENERATIVE-FIX- DIVERGENCE guard). - .changeset/quick-hawks-bark.md: Added fragment. Companion to #1945 (tracer bullets). * chore(#1949): regen agent-size baseline + install-tree goldens Documented baseline regenerations required by the feat(#1949) prose changes (RULESET.AGENT_SIZE_BUDGET + golden-install-parity): - npm run size:baseline — locks in the new gsd-executor.md size (+1050 bytes: the precondition-check step 0 block). gsd-planner.md is net smaller (-142 bytes: compressed two inline summary blocks whose full rules already lived in planner-antipatterns.md to make room for the slim <precondition> pointer). No hard-cap breach. - npm run gen:golden — pick up the new references/planner-preconditions.md + the two changed agent files across all 18 runtime install trees. Both regens are CI-mandated after intentional agent/reference changes; see CLAUDE.md 'RULESET.AGENT_SIZE_BUDGET' and the comments in tests/golden-install-parity.test.cjs. * fix(#1949): bound <precondition> checks to read-only (security review) Apply the security-review finding (LOW, isolated /security-review subagent): the executor's 'run the cheapest check' phrasing for a plan-author-controlled prose line was broader than ideal — a hostile plan author could craft a <precondition> whose 'cheapest check' is side-effecting (curl to an attacker host under the guise of verification, rm -rf before checking, secret emission). The risk is inherited from GSD's existing plan-trust model (<verify>, <action>, <done> already direct the executor to run arbitrary shell), so <precondition> does not materially expand it. But the new prose actively directs execution ('run the check') rather than passively consuming the element, so the bound is worth making explicit. Tightened across all four surfaces that describe the check shape: - agents/gsd-executor.md step 0: 'Verify with read-only checks only — file existence, env var presence (no value output), idempotent GET /health-style pings. Do NOT run commands with side effects (writes, network POSTs, secret emission) as the check; if a side-effecting check seems required, halt and surface via checkpoint instead.' - gsd-core/references/planner-preconditions.md Format section: same bound, plus the halt-and-surface escape hatch. - docs/reference/plan-md.md Preconditions section: mirrored. - CONTEXT.md Precondition glossary entry: mirrored. Regenerated agent-size baseline (executor grew 46186 -> 46440; still under the 49152 cap) and install-tree goldens. * chore(#1949): backfill changeset pr number 2422 Per CONTRIBUTING.md changeset workflow + feature-builder directive Step 8.7: backfill the placeholder pr:0 with the real PR number immediately after gh pr create returns. Avoids the fail_invalid_fragment gate. * fix(#1949): cite [#1949] on allow-test-rule exemption (ADR-456) CI's lint:ci runs lint-allow-test-rule-refs which per ADR-456 requires every // allow-test-rule: exemption on a NEW test file to carry an issue reference (#NNN or URL). My earlier push omitted it. Local 'npm run lint' (eslint) does NOT run this check — only 'npm run lint:ci' does. CLAUDE.md explicitly warns: 'lint:ci ≠ lint — CI runs lint:ci; a local pass is not the gate.' I should have run lint:ci before pushing; correcting now. Pattern matches the companion feature's test file: tests/tracer-bullet.test.cjs:1 // allow-test-rule: source-text-is-the-product [#1945]	2026-07-19 07:52:36 -04:00
Tom Boucher	8d2f8bcb23	fix(#2388): gate shared requirement completion on sibling plans, revert on gaps (#2424) ... * fix(#2388): gate shared-ID requirement marking and revert on gaps_found Adds requirements.ready-ids (execute-plan.md's update_requirements step) so a requirement ID declared by multiple plans in a phase only marks Complete once every declaring plan has produced a SUMMARY.md, and requirements.revert-phase (execute-phase.md's gaps_found branch) so a gaps_found verdict reverts the phase's own prematurely-Complete IDs before the gap report renders. Single-plan IDs still mark immediately. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * chore(#2388): regenerate fixtures + lint gate-prep * fix(#2388): repair failing tests after gate verification * chore(#2388): add changeset (#2424) --------- Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>	2026-07-19 07:52:17 -04:00
Tom Boucher	dd5a2211c9	enhance(#1964): semantic knowledge-base recall via MemPalace (keyword fallback) (#2416) ... * test(#1964): add failing-first semantic-recall contract tests Epic #1957 Phase 3C (final). Source-text-is-the-product contract tests: semantic recall via MemPalace (top-k meaning-similar prior resolutions, catches same-root-cause/different-wording cases), indexing resolved sessions at archive, graceful degradation to keyword matching when MemPalace is absent, knowledge-base.md stays the durable plain-text source of truth, agent Phase 0 / Matching Logic is semantic-first (the stale 'keyword overlap, not semantic similarity' claim must go), and no new embedding/vector infra (reuse MemPalace). Failing-first: reference, the Matching Logic reframe, the Phase 0 consolidation, and the archive indexing step do not yet exist. * feat(#1964): semantic knowledge-base recall via MemPalace (keyword fallback) Epic #1957 Phase 3C (FINAL). Replaces keyword-overlap matching with semantic recall: at Phase 0 the debugger queries MemPalace with the current symptoms and surfaces the top-k meaning-similar prior resolutions, catching the same-root-cause/different-wording cases keyword overlap missed (the self-noted 'keyword overlap, not semantic similarity' limitation). Resolved sessions are indexed into MemPalace at archive (symptoms + root_cause(s) + fix + recurrence guard). knowledge-base.md remains the durable plain-text source of truth; when MemPalace is absent the debugger falls back to keyword-overlap matching (logged, never a silent skip). No new embedding/vector infrastructure — MemPalace is reused. Size-neutral agent edits: the Matching Logic section reframed (keyword-only -> semantic-first + keyword-fallback + @-include); Phase 0's three keyword bullets consolidated into one semantic-first bullet; one MemPalace-indexing step added at archive. Agent at 57222 B (122 B headroom — final phase). Full rules in gsd-core/references/debugger-semantic-recall.md. INVENTORY + manifest + agent-size baseline + install-parity goldens + AGENTS.md updated. * fix(#1964): address orthogonal review (invocation mechanism, index Resolution-not-symptoms + redaction, fallback detail) - HIGH: the 'query MemPalace' instruction was WHAT-level only; the agent has no MCP tools. Added an Invocation section naming the Bash CLI (mempalace search --wing <wing>) + MCP-when-registered + wing resolution (config.mempalace.wing -> project_code -> project dir), matching every other MemPalace integration. Without this the feature silently degraded to keyword matching even when MemPalace was present. - MEDIUM (security x2): index the agent-authored Resolution summary (root_cause + fix + recurrence_guard), NOT raw user-supplied Symptoms — excludes attacker-controlled prose from the cross-session index AND reduces secret/PII leakage. Redact secret-shaped values before indexing. Stated the write order (KB append + commit MUST succeed before indexing). - LOW: restored 'identifiers' + 'case-insensitive' to the keyword fallback; added a test asserting the fallback mechanics survived the Phase 0 consolidation (Error patterns field, 2+ token overlap, identifiers, case-insensitive). * chore(#1964): ratchet agent-size baseline downward (leaner archive bullet shrank gsd-debugger.md 57222->57197) * chore(#1964): backfill changeset pr number (PR #2416)	2026-07-18 19:01:04 -04:00
Tom Boucher	c67f301867	feat(#1963): emit blameless-postmortem Prevention block at resolution (#2410) ... * test(#1963): add failing-first prevention/postmortem contract tests Epic #1957 Phase 3B. Source-text-is-the-product contract tests: blameless 5-Whys that BRANCHES per Phase 2A RCA (not a single-cause chain; treats agent error as 'why was that possible?'), the 'why wasn't this caught?' question, the recurrence-guard taxonomy (regression test / assertion / lint rule / KB pattern), the KB-entry why_not_caught + recurrence_guard fields with backward compat, the session-manager prevention summary line, and the Zawinski scope-boundary (a block, not a subsystem). Failing-first: reference, archive_session edit, KB schema extension, and session-manager summary do not yet exist. * feat(#1963): emit blameless-postmortem Prevention block at resolution Epic #1957 Phase 3B. At archive_session the debugger now produces a Prevention block with three blame-free components: a branching 5-Whys causal chain (branches per Phase 2A RCA, not a single chain; 'agent error' prompts 'why was that possible?', never blame), a 'why wasn't this caught?' answer naming the missed gate (test/typecheck/lint/review/verify), and a concrete recurrence guard (regression test / assertion / lint rule / KB pattern). The knowledge-base entry gains two structured fields (why_not_caught + recurrence_guard) so future Phase-0 recall surfaces the prior prevention, not just the prior fix. Additive: old entries without the fields still load. The session-manager compact summary surfaces a one-line prevention summary. Full rules extracted to gsd-core/references/debugger-prevention.md (slim archive_session step + 2 KB fields kept in the agent). INVENTORY + manifest + agent-size baseline + install-parity goldens + AGENTS.md updated. * fix(#1963): address orthogonal review (CRITICAL append-template drift + Phase-0 consumption + parity test) - CRITICAL: the archive_session KB append template omitted Why not caught + Recurrence guard (only the Entry Format had them) — the feature's core deliverable silently did not happen. Added both fields to the append template the agent actually follows (nearest-instruction wins). - HIGH: Phase 0 (KB read) only surfaced root_cause + fix; the new fields were dead data. Extended the Phase 0 Evidence line to consume why_not_caught + recurrence_guard when present (absent on old entries — backward compat holds). - MEDIUM: added a cross-section parity test (every Entry-Format field must also appear in the append template — the guard that would have caught the Critical) + a Phase-0-consumption assertion. - MEDIUM: the 'branches per Phase 2A' claim is now wired — reuses reasoning_checkpoint.candidate_causes across the four categories. - MEDIUM: recurrence-guard taxonomy gains type refinement + config-default change; LOW: added 'build' gate to both surfaces for parity. - NIT: compact-summary fallback shape ('no gate existed'); verify the guard artifact exists before recording it. * test(#1963): anchor Phase-0 consumption test on the specific heading The regex /Phase 0[\s\S]{0,1200}/ matched the first 'Phase 0' in the file (in knowledge_base_protocol prose), not the Phase 0 block in investigation_loop. Anchor on '**Phase 0: Check knowledge base**' and widen to 1500 chars. * chore(#1963): backfill changeset pr number (PR #2410)	2026-07-18 17:38:54 -04:00
Tom Boucher	36a311c5bb	enhance(#1962): harden regression tests (PBT shrinking + oracle classification + boundaries) (#2409) ... * test(#1962): add failing-first repro-hardening contract tests Epic #1957 Phase 3A. Source-text-is-the-product contract tests: PBT shrinking (fast-check/Hypothesis, minimized seed, manual-minimization degradation), the four oracle types (specified/derived/metamorphic/implicit with implicit flagged weakest), boundary neighbors (off-by-one/min-max/empty-singleton tied to the equivalence class), oracle_type in DEBUG Resolution, and the Phase 1A tie-in (minimized seed + real oracle => the mutation guardrail bites). Failing-first: reference, agent cross-refs, and template field do not yet exist. * feat(#1962): harden regression tests (PBT shrinking + oracle classification + boundaries) Epic #1957 Phase 3A. Extends Minimal Reproduction (shrinking) and Test-First Debugging (oracle classification + boundary neighbors): - Shrinking: wrap an input-space failing input in a property (fast-check JS/TS, Hypothesis Python) and store the MINIMIZED counterexample as the regression seed; degrade to manual minimization when no PBT framework is present. - Oracle classification: state specified / derived (contract/model) / metamorphic / implicit (crash, weakest) before writing the assertion; record under Resolution.oracle_type; never default to implicit silently. - Boundary neighbors: off-by-one, min/max, empty/singleton around the fixed defect's equivalence class. Together they turn the regression test into a root-cause check — what the Phase 1A mutation guardrail needs to bite. Full rules extracted to gsd-core/references/ debugger-repro-hardening.md. INVENTORY + manifest + agent-size baseline + install-parity goldens + AGENTS.md + DEBUG template updated. * fix(#1962): address orthogonal review (bounding, provenance, oracle scope, sufficient-triple) - HIGH: added a 'Bound the property/shrink run' section (60s timeout, degrade- to-manual on timeout, do-not-raise-default-run-limits, argv-not-shell) — the gauntlet violation the sibling references already honored. - Medium: test-provenance caveat (the failing input often comes from the bug report — author the generator from a sanitized description, cross-ref debugger-fix-acceptance.md). - Medium: oracle scope note — the 4 types cover deterministic bugs; non- deterministic failures re-route to stability-stress per bug-taxonomy. - Medium: Phase 1A tie-in corrected — seed+oracle is necessary not sufficient; boundary neighbors close the adjacent-input escape; the sufficient triple is seed+oracle+neighbors. - Low: preserve the original noisy repro as a secondary reference; operationalize 'equivalence class' (the predicate the fix draws). Nit: degradation reworded. * chore(#1962): backfill changeset pr number (PR #2409) --------- Co-authored-by: sim <sim@local>	2026-07-18 15:46:42 -04:00
Tom Boucher	6baa2a8182	feat(#1961): add bug-taxonomy classification + strategy routing to gsd-debugger (#2407) ... * test(#1961): add failing-first bug-taxonomy routing contract tests Epic #1957 Phase 2B. Source-text-is-the-product contract tests (3 taxonomy classes, explicit class->technique routing table, Bohrbug->repro+SBFL+bisect, Heisenbug->record-replay/stability+SKIP-SBFL, Concurrency->atomicity/order/ deadlock checklist, bug_class in DEBUG Current Focus, supersede-not-append) plus a routing-table specification object pinning the documented decisions (SBFL forbidden on Heisenbug is the load-bearing 1B/2B seam). Failing-first: reference, Phase 1.75, and routing-table reframe do not yet exist. * feat(#1961): add bug-taxonomy classification + strategy routing to gsd-debugger Epic #1957 Phase 2B (reliability-critical). Adds Phase 1.75: classify the failure as Bohrbug / Heisenbug-Mandelbug / Concurrency, then route the investigation technique via an explicit class->technique table (Kernighan: no opaque heuristic). Bohrbug -> reproduction + SBFL (Phase 1.25) + git bisect; Heisenbug/Mandelbug -> record-replay (rr) + stability-stress + statistical sampling, with SBFL explicitly SKIPPED (a flaky spectrum poisons the Ochiai ranking — the load-bearing 1B/2B seam); Concurrency -> the atomicity/order/deadlock checklist first. Reframes (supersedes, not appends — Zawinski) the flat 'Technique Selection by situation' table into a class-routed table; the 11 techniques remain as routed targets. bug_class recorded in Current Focus (DEBUG template); common-bug- patterns catalog cross-referenced to the taxonomy. Full rules extracted to gsd-core/references/debugger-bug-taxonomy.md. INVENTORY + manifest + agent-size baseline + install-parity goldens + AGENTS.md updated. * fix(#1961): address orthogonal review (phase-name drift, General lane, revoke framing, row-scoped tests, bounding) - HIGH: reference said 'Phase 1B' (epic shorthand); corrected to the deployed 'Phase 1.25' (matches the agent + SBFL reference). - HIGH: 6 of 11 techniques (Rubber duck, Delta, Working backwards, Differential, Comment-out, Follow-the-indirection) were orphaned by the situation-table reframe. Added a 'General (any class, situation-cued)' lane to BOTH the reference routing table and the agent's Technique Selection table that re-homes them — supersede-not-append now holds. - MEDIUM: the SBFL-skip is structurally retroactive (Phase 1.25 runs before Phase 1.75 classification), so reframed the table column from 'Do NOT use' to 'Revoke if already run' + an explicit 'retroactive revocation, not proactive skip' note stating the ordering honestly. - MEDIUM: contract tests are now row-scoped (parse the table by class, assert per-row) instead of presence-only; added a guard that the previously- orphaned techniques now have a General-lane route. - LOW: pinned the canonical bug_class value form (lowercase-kebab: bohrbug|heisenbug-mandelbug|concurrency; prose may use title-case). - NIT: added a 'Bound the Heisenbug-chase runs' note (rr/stability/sampling timeouts) per the unbounded-subprocess gauntlet. * chore(#1961): backfill changeset pr number (PR #2407)	2026-07-18 14:42:29 -04:00
Tom Boucher	f8b16d1874	enhance(#1960): add RCA branching (fishbone + AND-gate) to gsd-debugger (#2405) ... * test(#1960): add failing-first RCA-branching contract + schema-invariant tests Epic #1957 Phase 2A. Source-text-is-the-product contract tests (fishbone >=2 categories, AND-gate, multi-cause root_cause, backward compat, reasoning checkpoint candidate_causes+and_gate fields, debugger-philosophy single-cause note, DEBUG template) plus behavioral schema-invariant checks on two fixtures: two contributing causes (AND-gate yes) -> both recorded; single-cause (AND-gate no) -> one root_cause, identical to today. Failing-first: reference, agent edits, and template note do not yet exist. * feat(#1960): add RCA branching (fishbone + AND-gate) to gsd-debugger Epic #1957 Phase 2A. Guards against 5-Whys single-cause bias: before committing root_cause, the debugger enumerates candidate causes across >=2 Ishikawa categories (code/config/environment/data) and explicitly answers an AND-gate question. When the AND-gate fires, every contributing cause is recorded, so a multi-cause fix no longer recurs via the unaddressed second cause. Resolution.root_cause may hold one OR a small set (additive; single-cause sessions are byte-identical to today). The Structured Reasoning Checkpoint gains candidate_causes + and_gate fields; debugger-philosophy.md adds the single-cause-bias trap. Full rules extracted to gsd-core/references/debugger-rca-branching.md (slim Phase 2 routing + 2 checkpoint fields kept in the agent). INVENTORY + manifest + agent-size baseline + install-parity goldens + AGENTS.md + DEBUG template updated. * fix(#1960): address orthogonal review (AND-gate self-consistency, parity guard, narrowed claim, ripples) - Reference: the collapse rule now enforces AND-gate self-consistency — and_gate=yes with a single confirmed cause is flagged as incomplete (return to Phase 3); a race/timing note clarifies such bugs bridge categories; the 'byte-identical' backward-compat claim narrowed to 'root_cause shape unchanged; reasoning_checkpoint gains 2 fields in every session'. - DEBUG.md: stale 'five-field' mirror prose -> seven-field (parallel-surface drift the reviewer flagged); new debug-session-management parity test pins the field-count claim to the gsd-debugger.md YAML keys (CRLF-safe). - Scalar-assuming consumers of set-valued root_cause updated: session-manager compact summaries (319/332), diagnose-only return (1062), archive entry (1216), ROOT CAUSE FOUND return (1322). - Test: added the AND-gate-yes/single-cause invariant + fixture; rephrased the fixture describe block honestly as a schema-invariant specification. - Phase 2 bullet phrasing clarified ('at hypothesis formation, before the Phase 4 commit'). * test(#1960): parity regex accepts word-form count ('seven-field' or '7-field') * test(#1960): parity regex counts array-valued YAML keys (no inline value) * chore(#1960): backfill changeset pr number (PR #2405)	2026-07-18 13:42:58 -04:00
Tom Boucher	56a5c6404c	feat(#1959): add spectrum-based fault localization (Ochiai) pre-filter to gsd-debugger (#2403) ... * test(#1959): add failing-first SBFL contract + Ochiai correctness tests Epic #1957 Phase 1B. Source-text-is-the-product contract tests (Ochiai formula documented, Tarantula fallback, top-N seeding, no-coverage skip logged, ranking->Evidence, Bohrbug gating) plus a behavioral Ochiai formula-correctness section: bound [0,1], max-score invariant, a known-fault fixture proving the fault ranks #1 (criterion 2), clean degradation on zero failing tests, and two fast-check properties. Failing-first: reference file and agent routing do not yet exist. * feat(#1959): add spectrum-based fault localization (Ochiai) pre-filter to gsd-debugger Epic #1957 Phase 1B. When a runnable test suite with per-test coverage exists (>=1 failing AND >=1 passing test), the debugger computes an Ochiai suspiciousness ranking over the coverage spectrum and seeds the top-N suspicious locations into Evidence as first-class hypothesis candidates, narrowing the search space deterministically before LLM reasoning. Tarantula documented as fallback. Degrades cleanly (logged, never silent) when there is no test suite, no failing tests, or no per-test coverage, and is explicitly not trusted on flaky/Heisenbug spectra (pairs with Phase 2B bug-taxonomy). Full rules extracted to gsd-core/references/debugger-sbfl.md (slim Phase 1.25 routing kept in the agent to respect the size cap). No new coverage framework — reuses the project's existing test/coverage runner. INVENTORY + manifest + agent-size baseline + install-parity goldens + AGENTS.md updated. * test(#1959): bound property generators to valid coverage counts The [0,1] property generated failedExec independently of totalFailed, but Ochiai's score is only bounded by 1 under the coverage invariant failedExec <= totalFailed (a failing test that executed s is one of the totalFailed failing tests). Out-of-domain inputs (failedExec=100, totalFailed=5) make the formula correctly return >1. Bound failedExec by totalFailed via fc.chain so the property tests the real domain. Also cleaned up the ranking property (removed dead code). * fix(#1959): address orthogonal review (monotonicity property, degradation row, coverage bounding) - Replace vacuous ranking property (true-by-sort-construction) with a non-trivial monotonicity property: holding totalFailed + passedExec fixed, ochiai is non-decreasing in failedExec. An inverted formula would fail it. - Add the missing 'no passing tests' degradation row (preconditions require >=1 passing test; Tarantula would divide by totalPassed=0). - Bound the coverage subprocess (CLAUDE.md gauntlet): cap the coverage run, degrade-to-skip on timeout, never hang the debug session. - Reword 'discard the ranking' -> 'mark the Evidence entry as revoked (do not delete)' per Kernighan auditability. * test(#1959): bound monotonicity-property generator to valid coverage (failedExecA <= totalFailed) * chore(#1959): backfill changeset pr number (PR #2403)	2026-07-18 07:55:28 -04:00
Tom Boucher	5e52350736	feat(#1958): add multi-signal fix-acceptance guardrail to gsd-debugger (#2396) ... * test(#1958): add failing-first guardrail contract tests Epic #1957 Phase 1A. Adds source-text-is-the-product tests asserting the 5-signal fix-acceptance guardrail contract (target test, mutation check, no-op/deletion detector, adjacent tests, revert-and-reconfirm), graceful degradation, FIX REJECTED BY GUARDRAIL return path, per-signal debug-file recording, and subprocess bounding. Failing-first: reference file and agent sections do not yet exist. * feat(#1958): add multi-signal fix-acceptance guardrail to gsd-debugger Epic #1957 Phase 1A. Prevents accepting a fix that merely greens the test (Goodhart defense / APR overfitting). Adds a 5-signal gate run before fix acceptance: target test, mutation check (Stryker), no-op/behavior-deleting detector, adjacent/held-out tests, revert-and-reconfirm. Degrades gracefully when Stryker or a test suite is absent (each skip logged, never a silent pass), records per-signal results under Resolution.verification, and returns a FIX REJECTED BY GUARDRAIL outcome the session-manager surfaces for revise / accept-as-debt / abandon. Full rules extracted to gsd-core/references/debugger-fix-acceptance.md (slim routing kept in the agent to respect the agent-size cap). Debug template + INVENTORY + manifest + agent-size baseline + AGENTS.md updated. * test(#1958): correct newline-tolerant assertion + regen install-parity goldens The revert-and-reconfirm assertion collapsed whitespace before matching so markdown line-wrapping does not break it. Regenerated the golden-install-parity and install-tree fixtures (npm run gen:golden) to absorb the intentional gsd-debugger.md / gsd-debug-session-manager.md / DEBUG.md / new reference-file changes to the installed artifact tree. * fix(#1958): tighten guardrail per orthogonal review Addresses the isolated reviewer's findings: - signal 5 now states its recorded-repro dependency and routes the no-repro case to the degradation row; revert mechanism specified (git stash / git revert -n); minimality flag tied to diff structure, not revert-ability. - bounded-subprocesses section now bounds the git subprocess (5-30s) too, requires argv-array argument passing, and scopes Stryker to the driving regression test (a mutant killed only by a non-driving test is a finding). - new test-provenance (security) clause: the driving test must be agent-authored; bug-report repro scripts are DATA, never executed verbatim. - tightened 3 contract assertions to bind to specific clauses (guardrail_verdict field, deletion-reject-unless-RCA, 60s+git bounding). - Goodhart framing softened to 'partially-independent'; DEBUG.md template verification field notes the nested map shape. * chore(#1958): backfill changeset pr number (PR #2396) * fix(#1958): add issue ref to allow-test-rule annotation (ADR-456) CI lint-allow-test-rule-refs requires every allow-test-rule exemption to carry a 'see #NNN' issue ref per ADR-456. The new test file's annotation lacked it; this adds (see #1958).	2026-07-18 00:39:28 -04:00
Tom Boucher	315d94f6d4	feat(#1945): tracer-first planning default + executor feedback gate (#2294) ... * feat(#1945): tracer-first planning default + executor feedback gate Make "thin end-to-end slice first, verify, then expand" the default planning + execution discipline instead of the opt-in --mvp mode. - gsd-planner: first-class `type="tracer"` task; every plan LEADS with one production-quality end-to-end tracer slice by default; --no-tracer restores horizontal layers; --mvp/--tdd compose on top. - gsd-executor + execute-plan: post-tracer feedback gate — autonomous runs halt-on-fail before expansion, interactive runs emit checkpoint:human-verify after the tracer. - --no-tracer flag wired through plan-phase workflow/command/help/skill. - CONTEXT.md glossary defines tracer bullet vs prototype; docs + references reconciled. - tests/tracer-bullet.test.cjs: prose-contract + behavioral (verify plan-structure accepts tracer) coverage. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * chore(#1945): backfill changeset PR number to 2294 --------- Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>	2026-07-15 09:41:36 -04:00
Adnan	3592697bed	fix(#2107): orchestrator honors gate="blocking-human" checkpoints in auto-mode (#2113) ... * fix(execute-phase): honor gate="blocking-human" in auto-mode checkpoint handling The package-legitimacy gate (#2827) spans two layers. gsd-executor refuses to auto-approve a gate="blocking-human" checkpoint and escalates it so a human can vet the package. execute-phase's checkpoint_handling step then dispatched purely on checkpoint *type* and never read gate -- so under --auto/--chain it auto-approved the checkpoint the executor had just refused to auto-approve. Net effect: the slopsquatting defence was inert in exactly the unattended mode where it matters. An [ASSUMED]/[SUS] package reached install with no human ever seeing the prompt. - gsd-core/workflows/execute-phase.md: carve out gate="blocking-human" (and the package-legitimacy what-built markers) ahead of every auto-mode branch. - gsd-core/references/checkpoints.md: document the gate attribute and its two values. blocking-human previously appeared nowhere outside gsd-executor.md, so no planner had a documented way to author a non-auto-approvable checkpoint. - tests/package-legitimacy-gate.test.cjs: the existing regression test asserted the executor half only, which is why it stayed green while the gate was open. Now asserts the orchestrator half too. * chore(changeset): link to issue #2107 * chore(changeset): backfill PR number 2113 Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01JNR8m2pv5U7ubn4iiXVrMa * test(#2107): refresh golden-install-parity hashes for edited gsd-core files The golden fixtures pin content hashes for gsd-core/references/checkpoints.md and gsd-core/workflows/execute-phase.md, both edited by this fix. Regenerated via UPDATE_GOLDEN=1; only those two keys change across all 17 runtime fixtures. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01JNR8m2pv5U7ubn4iiXVrMa * fix(#2107): keep the carve-out inside the ADR-857 host-loop budget The ADR-857 phase-6 ratchet pins execute-phase.md below 93600 LF bytes so optional-feature logic keeps migrating out of the host loop. The carve-out first landed 623 bytes over that ceiling. Move the two-layer rationale (why gsd-executor escalates these checkpoints) into references/checkpoints.md, where the gate is now documented, and reduce the workflow to the operative rule. execute-phase.md is 93589 bytes, under the ceiling; the gate token and both <what-built> marker strings are kept because the orchestrator matches on them. Refresh the two baselines the edit invalidates: golden-install-parity fixtures (only the checkpoints.md and execute-phase.md hashes move) and workflow-size-baseline.json (one line). The ADR-857 ceiling itself is untouched. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01JNR8m2pv5U7ubn4iiXVrMa * fix(#2107): executor honors blocking-human on the decision branch + gate transport Review found the fix incomplete one layer down. Two executor-layer gaps: 1. Blocker — agents/gsd-executor.md auto-mode dispatch gated checkpoint:human-verify on gate="blocking-human" but the checkpoint:decision branch below auto-selected the first option with no gate check. The executor resolves a decision itself (auto-selects and continues) without returning it, so the orchestrator carve-out never runs for it. A planner following the new checkpoints.md rule 6 ("gate a decision whose default would be wrong to assume") would have it silently auto-selected under --auto/--chain — the exact #2107 harm, one checkpoint type over. The decision branch now STOPs and returns for an explicit human decision when gate="blocking-human". 2. Major (transport) — checkpoint_return_format carried no field conveying the gate to the freshly-spawned orchestrator, so recognition of the proactive pre-install checkpoint rested on freeform prose. Added a **Gate:** field to the return format and re-pointed the execute-phase carve-out at it ("If the returned Gate: is blocking-human"). Net byte-negative: execute-phase.md drops 93589 -> 93583, widening ADR-857 headroom from 11 to 17 bytes. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * test(#2107): cover decision carve-out + gate transport, de-vacuum conditional tests - New: 'auto mode does not auto-select a blocking-human decision checkpoint' asserts the executor decision branch STOPs on blocking-human. Verified red on the pre-fix executor (2 fail), green with the fix (27 pass). - New: 'checkpoint_return_format transports the gate ...' asserts the **Gate:** field carries blocking-human across the executor->orchestrator boundary. - New: 'auto-select rule for decision is conditional' — orchestrator-side mirror of the human-verify conditional test, for the execute-phase decision branch. - Fix vacuous test: both conditional tests now assert the anchor matched (length > 0) before iterating, so anchor drift can no longer pass with zero assertions. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * test(#2107): refresh golden + size baselines for executor + execute-phase edits Regenerated via UPDATE_GOLDEN=1 and update-size-baseline.cjs. Only the gsd-executor.md and gsd-core/workflows/execute-phase.md hashes move across the runtime fixtures (35 ins / 35 del, no keys added or removed); checkpoints.md is unchanged this round. Size baselines: gsd-executor.md 43607 -> 43973, execute-phase.md 93589 -> 93583 (still under the ADR-857 ceiling). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com> Co-authored-by: Tom Boucher <trekkie@nomorestars.com>	2026-07-13 13:43:29 -04:00
Dave	29120ea0aa	docs(#1867): ui-consideration-probe reference + docs-parity test (ADPT-02) ... Reference doc mirrors edge-probe.md structure but links rather than re-argues; states the MIXED-axis boundary (closed compiled shape-rooted subset here; open UX subset - real-time/offline, a11y depth, i18n/RTL - prose-owned in domain-probes.md). Docs-parity test pins doc taxonomy ids == code UI_TAXONOMY ids and asserts disjointness from domain-probes.md topics (ADR-456 runtime-contract exemption, see #1867).	2026-07-10 14:40:03 -04:00
Tom Boucher	1ee00320c7	Merge branch 'next' into fix/2072-thread-model-into-assumptions-and-review-spawns	2026-07-07 22:03:47 -04:00
Tom Boucher	4483300253	fix(#2072): thread resolved model into routed-agent spawns (assumptions-analyzer, code-reviewer, code-fixer) ... model_overrides / models.<phaseType> were silently inert for gsd-assumptions-analyzer, gsd-code-reviewer, and gsd-code-fixer on Claude Code: resolveModelInternal honors them, but the workflows spawned these agents with no model= param, so the resolved value never reached the Agent tool and the agents inherited the session model — no warning. Fix — thread each agent's resolved model at every spawn site (the established plan-phase pattern; the architecture-consistent Claude mechanism, since 13 other agents already thread their model): - discuss-phase-assumptions.md: `resolve-model gsd-assumptions-analyzer --raw` → ANALYZER_MODEL, threaded. - code-review.md + code-review-fix.md (re-review): `resolve-model gsd-code-reviewer --raw` → REVIEWER_MODEL, threaded. - code-review-fix.md (both fixer spawns): `resolve-model gsd-code-fixer --raw` → FIXER_MODEL, threaded (same silently-inert bug, same file — folded in per review). - quick.md review step: was reusing `{executor_model}` for gsd-code-reviewer (so the reviewer's own override was ignored); init.quick now resolves `reviewer_model` (gsd-code-reviewer) and the spawn threads it. resolve-model --raw returns the bare model string (resolve-execution --raw would return effort — wrong). The resolver maps these agents to phaseType discuss / verification / execution, so models.<phaseType> apply too. Scope: the three agents reachable from the two issue-named workflows + quick.md. The wider systemic class (other agents in UNTOUCHED workflows with the same pattern) stays documented on the issue for a maintainer-scoped structural decision (thread-at-source vs embed-at-install like #2256), not widened here. Docs: the stale "discuss — reserved, no subagent today" model-profile tables now list gsd-assumptions-analyzer and the verification row includes gsd-code-reviewer, across the English docs, the shipped gsd-core/references/model-profiles.md reference, and the ja-JP / zh-CN / ko-KR / pt-BR locale mirrors. Tests: - tests/model-resolver.test.cjs: #2072 acceptance — model_overrides and models.discuss/verification/execution resolve for all three agents. - tests/model-routing-spawn-threading.test.cjs: every spawn of the three agents threads a resolved model (fails pre-fix); a header-precise parity guard fails the suite if a new un-threaded spawn of any of them regresses. All 16 golden-install-parity fixtures + the workflow size baseline regenerated for the changed shipped files (4 workflows + the reference doc); bin/lib is excluded from parity. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>	2026-07-07 21:22:48 -04:00
Tom Boucher	6addeccd19	feat(ai-integration): API-coverage verify:pre gate (#1562) ... Full API Coverage by Default — Opt Out, Never Opt In. A phase that integrates an external API/SDK/service can no longer seal without a decided coverage matrix. - src/api-coverage.cts: deterministic detector (compound verb+noun signal + <Service> API/SDK surface; stopword-guarded; strips fenced code) + matrix parse/validate/render with field-length caps. - check api-coverage.verify-pre: blocking seal-time gate; phase arg resolved as a token under .planning/phases/ only (traversal-neutralized); validates COVERAGE.md or blocks iff a strong integration signal is detected and no matrix exists; fail-closed when phases tree exists but phase unresolvable. - capabilities/ai-integration: workflow.api_coverage_gate config key (default true), plan:pre contribution, blocking verify:pre gate. Data-driven. - gsd-core/workflows/verify-work.md: generic verify:pre gate dispatch. - Tests: detector FP/FN + matrix validation + fast-check bijection; gate e2e. Code+security review findings fixed (stopword FP, scope containment, pipe/cap rejection, prompt-injection message hygiene). - Regenerated registry/matrix/loop-host-contract/goldens/baseline + docs. Closes #1562	2026-07-07 15:11:12 -04:00
Tom Boucher	ea4378063f	Merge branch 'next' into feat/1820-specless-predicate-rail	2026-07-07 07:49:07 -04:00
Tom Boucher	080bacdb4b	Merge branch 'next' into codex/gsd-onboard	2026-07-06 22:43:32 -04:00
Dave	7ef834cabc	feat(#1820): spec-optional predicate rail — author probe predicates into must_haves when SPEC omits them	2026-07-06 14:40:38 -04:00
Tom Boucher	f272983d90	fix(#2019): planning-config.md global-learnings path ~/.gsd/learnings → ~/.gsd/knowledge (#2026) ... * fix(#2019): planning-config.md global-learnings path ~/.gsd/learnings → ~/.gsd/knowledge The features.global_learnings row pointed at ~/.gsd/learnings/ but the implementation (src/learnings.cts, execute-phase.md) uses ~/.gsd/knowledge/. Docs now match the code. Closes #2019 * docs(#2019): backfill changeset pr 2026	2026-07-05 15:59:36 -04:00
Codesmith	425d2f2c8b	fix(#1990): keep onboard resolver delegation in sync with canonical launcher ... onboard.md delegates the gsd_run preamble to gsd-core/references/gsd-run-resolver.md via @-include, but two guards regressed once the canonical launcher snippet advanced to ${CLAUDE_CONFIG_DIR:-$HOME/.claude} (#2024): - runtime-launcher-parity (B2): the resolver reference still shipped the old $HOME/.claude arm. references/ is not covered by sync-runtime-launcher.cjs, so refresh the resolver bash block to be byte-equal to _runtime-launcher.snippet.sh. - /gsd:onboard command contract: sync-runtime-launcher.cjs had re-inlined the preamble into onboard.md (a delegating file). Teach the sync transform to strip-but-never-inline files that @-include the resolver, mirroring the exemption already in the parity test (B/B2). Regenerate golden install fixtures and the workflow size baseline for the smaller onboard.md. Co-authored-by: Codesmith <codesmith-bot@users.noreply.github.com>	2026-07-05 19:16:10 +00:00
Jeremy McSpadden	a5298c1fc0	refactor: project onboard routing in init	2026-07-05 19:16:08 +00:00
Jeremy McSpadden	e5ef323b15	feat(#1787): add /gsd:next smart entry workflow (#1798) ... * docs: design spec for /gsd smart-entry command Hybrid approach porting gsd-pi's smart-entry wizard to gsd-core: deterministic classifier (gsd-tools smart-entry --json) + markdown command/workflow with AskUserQuestion + --text fallback. Routing-first ('what now?' menu), 10 situations redesigned for gsd-core's phase loop. * feat: add /gsd-start smart-entry command State-aware front door adapted from gsd-pi's smart-entry wizard, redesigned for gsd-core's markdown-first, multi-runtime architecture. - src/smart-entry.cts: deterministic situation classifier (no-project, paused, blocked, verify-failed, needs-first-phase, planning, executing, verify-pending, idle-stranded, complete, unknown). Reads STATE.md, ROADMAP.md, git, and verify signals; emits JSON the workflow consumes. - gsd-tools.cjs: wire case + help listing. - commands/gsd/start.md + gsd-core/workflows/gsd.md: thin markdown dispatcher presenting an AskUserQuestion menu (with --text fallback for non-Claude runtimes) and dispatching to existing commands. Falls back to /gsd:progress if detection is unavailable. - help.md: document /gsd:start (parity with bug-2954). - tests: smart-entry.unit.test.cjs (classifier behavior across all situations + priority + JSON shape) and gsd-workflow.structure.test.cjs (markdown-layer invariants + every emitted command resolves to a real slash command). Spec: docs/superpowers/specs/2026-06-27-gsd-smart-entry-design.md Note: command-contract (ADR-0002) requires a gsd:* prefix, so the bare /gsd from the spec surfaces as /gsd-start. * refactor: rename smart-entry command to /gsd:next Rename the command from /gsd:start to /gsd:next per feedback. The command file is now commands/gsd/next.md (name: gsd:next) and the backing workflow is gsd-core/workflows/smart-entry.md (named for the smart-entry classifier and gsd-tools smart-entry subcommand; does not collide with the existing workflows/next.md, which is the progress --next sub-workflow). help.md and the spec updated to match. All affected tests (188) pass; lint:ci clean. * fix: smart-entry reads real STATE.md schema (nested progress YAML + body Phase field) Codex review found the classifier misread this repo's own STATE.md: it looked only for scalar current_phase/total_phases frontmatter and body fields named 'Current Phase'/'Total Phases', but real STATE.md stores the phase as body 'Phase: N' and total_phases/percent under a nested 'progress:' YAML object. Both came back null, so active projects (e.g. this repo at Phase 3 / verifying) wrongly classified as needs-first-phase. - detectSignals now reads total_phases + percent from nested progress{} first, then scalar fm, then body; current_phase falls back to the body 'Phase:' field (parseProsePhaseField lineage). - Add regression tests against the real schema (nested progress YAML + body Phase field) covering verify-pending + executing situations. Verified against this repo: now classifies verify-pending (was needs-first-phase). Coverage 93.25% lines / 86.99% branches. * fix(workflow): tiered fallback when gsd-tools is broken (not just smart-entry) Live test exposed a self-defeating fallback: when smart-entry --json failed because gsd-tools itself was broken (missing markdown-sectionizer.cjs), the workflow fell back to /gsd:progress — which also depends on gsd-tools and would dead-end too. Replace the single /gsd:progress fallback with a tiered recovery: 1. Probe gsd_run state-snapshot. If it ALSO errors, the whole tool layer is down — read .planning/STATE.md directly with the Read tool and synthesize a minimal situation + actions menu so /gsd:next stays useful. Surface a rebuild hint. 2. Only if smart-entry alone is missing (older gsd-core), fall back to /gsd:progress as before. Matches the direct-read resilience the live agent already did by hand. * docs: add gsd-next skill surface * chore: trigger no-mistakes validation * no-mistakes(review): Fix smart-entry phase ordering * no-mistakes(review): Fix decimal smart-entry phase ordering * no-mistakes(test): Fix smart-entry next test contracts * no-mistakes(document): Docs synced for smart entry * chore: add changeset fragment for #1798 (/gsd:next smart-entry workflow) Co-authored-by: Codesmith <codesmith-bot@users.noreply.github.com> * fix: shorten next.md description and update golden install parity fixtures Co-authored-by: Codesmith <codesmith-bot@users.noreply.github.com> * fix: update /gsd-next refs to /gsd:next in docs and add Smart Entry topic alias Co-authored-by: Codesmith <codesmith-bot@users.noreply.github.com> * chore: trigger no-mistakes validation * fix: regenerate INVENTORY-MANIFEST.json for new /gsd-next files Full CI caught that adding commands/gsd/next.md + gsd-core/workflows/smart-entry.md left docs/INVENTORY-MANIFEST.json stale (not in the affected-test scope that no-mistakes' test gate runs, so it surfaced in CI). Regenerated via node scripts/gen-inventory-manifest.cjs --write; inventory-manifest-sync test now passes. * fix: add 'next' to core_loop cluster, update INVENTORY-MANIFEST, fix gates.md ref Co-authored-by: Codesmith <codesmith-bot@users.noreply.github.com> * fix: regenerate golden install parity fixtures for /gsd:next Full CI (shard 3/3) caught that adding commands/gsd/next.md + the smart-entry workflow/lib made the per-runtime golden install parity fixtures stale across all 16 runtimes. Regenerated via UPDATE_GOLDEN=1 node --test tests/golden-install-parity.test.cjs. All 16 fixtures + inventory-manifest-sync now pass. * Fix smart-entry verify-failed phase scoping and empty resolve shim step Scope detectVerifyFailed to STATE.md's current phase so leftover higher phase directories cannot force verify-failed routing. Move the gsd_run shim resolver into the workflow resolve step so agents define gsd_run before the detect step runs smart-entry. * fix: recapture golden fixtures with updated gates.md hash (/gsd:next) Co-authored-by: Codesmith <codesmith-bot@users.noreply.github.com> * fix: recapture all 16 golden fixtures with updated smart-entry.md hash Co-authored-by: Codesmith <codesmith-bot@users.noreply.github.com> * chore: regenerate fixtures + inventory manifest after rebase onto next Rebased onto next which adopted #1837 (package-version normalization to <VERSION> in golden-install-parity hashes). Recaptured the golden fixture that needed it (hermes), re-sorted INVENTORY-MANIFEST.json, and regenerated the gsd-next / ns-workflow skill descriptions to match the command surface. Co-authored-by: Codesmith <codesmith-bot@users.noreply.github.com> * refactor(#1787): delegate /gsd:next in-project advancement to gated /gsd:progress --next Reconciles the /gsd:next smart-entry front door with the existing /gsd:progress --next engine (davesienkowski review on PR #1798). The classifier previously recommended /gsd:execute-phase directly for the `executing` situation, bypassing workflows/next.md Route 0 (resume-incomplete-phase invariant, #160) and Gates 1-3 — reproducing the duplication that got the old flat /gsd-next removed (#3054), plus a correctness hazard (executing the recorded current phase while an earlier phase is silently incomplete). Now planning/executing/verify-pending recommend `/gsd:progress --next` (single gated engine); the specific command stays an explicit secondary. Off-path states (no-project, paused, blocked, verify-failed, idle-stranded, complete) keep direct recommendations — smart-entry's distinct value over --next. Adds docs/adr/1787-gsd-next-smart-entry.md and a regression test locking the delegation contract. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * docs(#1787): avoid literal /gsd-next token in ADR (bug-3054 guard) The repo-invariants #3054 guard bans the removed /gsd-next slash form in docs surfaces. Refer to the removed command as `gsd-next` (prose) — the historical reference is unchanged, just the banned token is dropped. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * chore: gitignore compiled host-integration-sdk + handshake-serialized .cjs Pre-existing gap from #1683: these two src/*.cts modules compile to gsd-core/bin/lib/*.cjs but were omitted from the per-file ignore list, so `npm run build`/`npm test` left them as untracked build artifacts (dirty tree + accidental-commit footgun). Adds them alongside their siblings (host-integration.cjs, mcp-server.cjs, …). Found while finishing #1798. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * test(#1787): lock per-situation action invariants for all 11 situations + ADR typo Adversarial-review follow-ups: - Add a test asserting every situation's action set has exactly one recommended action, 1-4 unique-id /gsd:* actions (previously the one-recommended/1-4 invariant was only sampled for 6 of 11 situations). - Fix ADR typo: /gsd-progress → /gsd:progress. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * fix(#1798): split oversized test chunks so a slow shard can't trip the per-chunk timeout Root-cause of the intermittent `full test (windows-latest, 22, shard 1/3)` failure. It was NOT a leaked handle (the runner's kill message guesses that, but --test-force-exit already exits leaks cleanly). Diagnosis: - Ran every shard-1/3 file WITHOUT --test-force-exit + a 45s kill-timer: zero hangs, zero leaks — every file self-exits. So no leaked handle / hang. - CI activity profile: output kept flowing (slowly) right up to the 600.0s kill — a dead hang would go silent. => pure slowness. - Per-file timing: install-minimal-hooks.test.cjs is a 4987-line / 250-case consolidation file doing dozens of real installs — 41s even on a fast Mac (much worse on the slow Windows I/O path), plus an install-heavy cluster. Mechanism: MAX_FILES_PER_CHUNK=180 packed the whole ~171-file shard into ONE `node --test` chunk, so the entire shard's wall-clock ran against a single 600s per-chunk backstop. On slow Windows runners that single chunk crossed 600s and was killed mid-run — an intermittent false-negative gate that also hits `next` directly. Fix: lower MAX_FILES_PER_CHUNK 180 -> 90 so each shard splits into ~2 chunks, each with its own fresh 600s budget and a fresh node process (also relieves per-process memory pressure). Verified locally: shard 1/3 now runs as chunk 1/2 (90 files) + chunk 2/2 (81 files), 5323 tests, 0 fail. Also made the timeout kill-message name slowness as a cause instead of asserting a leak, so the next debugger isn't sent hunting a nonexistent handle leak. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> --------- Co-authored-by: Codesmith <codesmith-bot@users.noreply.github.com> Co-authored-by: Cursor Agent <cursoragent@cursor.com> Co-authored-by: Tom Boucher <trekkie@nomorestars.com> Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>	2026-07-03 12:18:25 -04:00
Behruz Nassre Esfahani	8189d2f098	enhance(#1872): document Claude Code advisor inheritance in model profiles (#1922) ... * enhance(#1872): document Claude Code advisor inheritance in model profiles Add an "Advisor Tool (Claude Code)" section to gsd-core/references/model-profiles.md: session-level advisor is inherited by all GSD subagents and composes with the per-agent profile/tier system, candidate executor/advisor pairings per profile (cost/quality/caching claims attributed to Anthropic's advisor-tool docs, not asserted as GSD behavior), when it is worth enabling vs not, and the session-level / no-per-agent-control constraint linking anthropics/claude-code#73072. Docs-only. Golden-install-parity fixtures recaptured for the edited reference file (hash-only, one line per runtime). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * chore(#1872): add changeset Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(#1872): use Documentation changeset type for docs-only change Changeset type was `Changed`, which triggers the docs-required lint (TRIGGERING_TYPES in scripts/lint-docs-required.cjs). This PR only touches gsd-core/references/model-profiles.md, so there is no docs/ file to pair with and docs-lint failed. `Documentation` is the correct type for a docs-only enhancement and is exempt from the trigger. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(#1872): put docs-exempt marker on its own line, revert to type Changed The prior fix (type: Documentation) was invalid — parse.cjs ALLOWED_TYPES is {Added, Changed, Deprecated, Removed, Fixed, Security}, so both changeset-lint and docs-lint failed with invalid_type. Real root cause of the original docs-lint failure: DOCS_EXEMPT_RE is anchored to match the `<!-- docs-exempt: ... -->` marker only on its own line, but the marker was tacked onto the end of the prose line, so it was never captured (docsExempt: null) and the triggering `Changed` fragment had no docs/ pairing -> fail_docs_missing. Fix: keep the valid `type: Changed` and move the marker to its own line. Verified locally: changeset-lint -> ok_fragment_present, docs-lint -> ok (own-line marker parses to ok_fragments_exempt). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Co-authored-by: Tom Boucher <trekkie@nomorestars.com>	2026-07-02 23:41:39 -04:00
Tom Boucher	3c13903dcd	feat(#1866): agent-side self-load of configured agent_skills ... Each of the 22 consumer agents now self-loads its configured agent_skills in its mandatory init step, so .planning/config.json agent_skills.<type> reaches the agent on every runtime — including Cursor and /gsd-autonomous, where Skill()-delegated workflow bash init did not reliably execute. - gsd-core/references/agent-skills-bootstrap.md: shared contract (query + Read + dedup guard that skips when <agent_skills> is already in the prompt, so Claude's orchestrator-side injection never doubles) - 22 agents/gsd-*.md: one self-load line naming the agent's own type - gsd-core/workflows/autonomous.md: note that delegated agents self-load - tests/agent-skills-bootstrap.test.cjs: regression + parity (CONSUMER_AGENTS bijection + fast-check property) — Generative-Fix-Divergence guard - docs: ADR-1866, CONFIGURATION dual-injection How It Works, INVENTORY row, Changed changeset Closes #1866	2026-07-01 20:09:01 -04:00
Rezolv	18995380ce	feat(#1154): honest verifier — abstain (insufficient_spec) on non-inferable backstop truths (#1738) ... * feat(verify-phase): honest verifier — abstain (insufficient_spec) on non-inferable backstop truths (#1154) Carry the edge-probe's existing `backstop` (non-inferable) tier through the plan-phase projection as a structured flat-scalar marker instead of a prose parenthetical, and make verify-phase abstain -> human_needed (never silent-pass) on a backstop truth it cannot confirm with explicit evidence. Truth-axis mirror of #644's prohibition judgment-tier (ADR-550 D4). Engine (deterministic, CI-tested per ADR-550 D5 — never the LLM verdict): - src/probe-core.cts: truthStatement/truthVerification normalizers, projectTruths (conservative serializer), dispositionForUnverifiableTruth (backstop+no-evidence -> unverified/flagged/insufficient_spec; backstop+evidence -> green; inferable -> green, the over-abstention guard). - src/roadmap.cts: coerceTruthToString now reads `statement` first so an object-form backstop truth is surfaced, not dropped (Hyrum backward-compat for truth-readers). Workflow/agent/docs: plan-phase emits the structured marker (flat scalar, ADR-550 #1278); verify-phase + gsd-verifier add the abstain arm; new references/honest-verifier.md; FEATURES/COMMANDS document insufficient_spec; ADR-550 amended (truth-axis D4 mirror). Decisions adopted (trek-e review): insufficient_spec feeds existing human_needed with a distinguishable reason (no new VERIFIER_STATUS); changeset Changed; round-trip parity test; abstain-on-unconfirmed-backstop regression test red-first. Implementation notes (deviations from the issue's proposed file list, verified live): - frontmatter.cts needs no change — its flat parser already round-trips object-form truths. - verify.cts needs no change — it grades artifacts/key_links structurally; truths are LLM-graded at the workflow layer, so consumption lives there + the deterministic helper. - No CJS<->SDK hand-sync — the SDK seam was retired (ADR-0174); src/*.cts is sole source. Regenerated artifacts: golden-install-parity fixtures, INVENTORY-MANIFEST, size baselines. * chore(#1154): add changeset (Changed) for honest verifier User-facing changelog fragment for #1738. Typed `Changed` (not `Added`) per trek-e review condition 3 — the verify behavior shifts for backstop-bearing specs (a confident silent `passed` becomes `human_needed`), which is user-visible even though the schema marker is additive. * docs(#1154): score-formula also excludes abstained insufficient_spec truths (review nit-1) trek-e review nit: the verify-phase score sentence said PRESENT_BEHAVIOR_UNVERIFIED truths were "the only ones excluded" from verified_truths. Post-#1154 an abstained `insufficient_spec` backstop truth is also excluded (it is not ✓ VERIFIED and routes to human_needed). Behavior was already correct; this tightens the wording. Regenerated golden-install-parity fixtures + workflow-size baseline for the touched verify-phase.md. (Nit-2 — a dedicated insufficient_spec_items frontmatter list — is intentionally not taken: the current design is ADR-550-D4-conformant, the abstain cause rides as a distinguishable report reason, and adding it would exceed the approved scope.) --------- Co-authored-by: Tom Boucher <trekkie@nomorestars.com>	2026-06-29 00:14:32 -04:00
Tom Boucher	b0d5ca3379	feat(#1517): support custom reviewer instances for /gsd:review (#1766) ... * feat(#1517): support custom reviewer instances for /gsd:review Add a bounded review.reviewer_instances config surface so one model-capable adapter (e.g. opencode) can run as several independent reviewer identities in a single /gsd:review pass. Instances participate only via review.default_reviewers, expand before built-in slugs, are available iff their cli is detected, and a non-matching entry is a hard error (typo must be loud). >=2 same-cli instances emit a shared-adapter caveat in REVIEWS.md. Default path with no instances is byte-for-byte unchanged. Single-source instance->cli resolution lives in resolveReviewerSelection / normalizeReviewerInstances (parity-locked in tests/review-reviewer-instances.test.cjs). cli validated against KNOWN_REVIEWER_SLUGS only (never arbitrary shell); model/agent opaque, never shell-interpolated. Closes #1517 * chore(#1517): backfill changeset pr:1766 --------- Co-authored-by: review-bot <review-bot@gsd>	2026-06-26 23:35:04 -04:00
Alex V.	a63684c222	enhance(#1577): WebFetch/WebSearch injection isolation + opt-in blocking (#1585) ... * fix(#1577): isolate WebFetch/WebSearch ingress + opt-in injection blocking Split A of #1573 (security-critical). Scans WebFetch/WebSearch output (the largest untrusted channel) in gsd-read-injection-scanner; shared untrusted-input-boundary reference @-included by the 8 ingest agents (randomized per-wrap delimiters, in-prompt self-scan guard, task-anchoring); opt-in security.injection_blocking (default advisory — non-breaking). arXiv: 2506.05739 (PPA), 2507.15219 (PromptArmor), 2504.20472 (Referencing), 2503.00061 (defense-in-depth). * fix(#1577): address review — honest blocking docs, config key, ADR, property test, revert localized - A1: rewrote the opt-in-blocking doc + Security changeset honestly — the PostToolUse hook is a circuit-breaker (halts the agent's next step), NOT a redactor; it does not scrub content already in the transcript. The prompt-level data/instruction boundary is the primary control. - A2: registered security.injection_blocking in the config schema + defaults manifests (default false) + an e2e config-roundtrip test; the dotted setter writes the nested shape the hook reads. - A3: reverted the 4 hand-edited localized security-model.md (canonical EN only, per convention). - A5: ADR-1577 (untrusted-input boundary + opt-in blocking; redaction-vs-circuit-breaker rationale). - A6: property test — scanner never crashes / only emits valid JSON on unicode/large/malformed input. - Also: inventory (untrusted-input-boundary.md) + agent-size baseline (8 ingest agents) + drift-guard matcher update (Read -> Read|WebFetch|WebSearch). A7 (content<20 early-exit) left as the noted pre-existing follow-up. * fix(#1577): allowlist untrusted-input-boundary.md in injection-scan CI gate The new reference quotes injection phrases ('ignore previous instructions', 'you are now…') as examples agents must NOT comply with, tripping the repo's own prompt-injection-scan.sh diff gate (the standalone 'security' CI job, red on HEAD). Allowlist it alongside the other security docs (security-model.md, TEST-EXAMPLES.md) that legitimately demonstrate injection patterns. The JS scanner test doesn't scan references/, so only the shell gate needed it. Verified: scan --diff origin/next -> 0 findings; scanner JS test 15/15. * fix(#1577): cover AC #2's gsd-ui-researcher + gsd-assumptions-analyzer trek-e Major 1: the @-included set dropped two AC #2 agents. Restore them so no named web-ingress agent is uncovered, keeping the two justified additions (gsd-ai-researcher, gsd-domain-researcher). Final set = AC's 8 + 2 = 10. - gsd-ui-researcher carries the full WebSearch/WebFetch + MCP-fetch toolset. - gsd-assumptions-analyzer reads 5-15 codebase source files (external/source- document ingress per the boundary), though it has no web tools. INGEST_AGENTS in the isolation test now asserts all 10; size baselines regenerated (+60 bytes each, both well under the DEFAULT cap); changeset reworded 8 -> 10. Verified: untrusted-input-isolation 14/14; agent-size-budget 39/39. * docs(#1577): document security.injection_blocking + boundary seam trek-e Major 2 + Minor: - docs/CONFIGURATION.md: add the top-level security.injection_blocking key to the Full Schema and a Security Settings subsection, distinguishing it from the workflow.security_* namespace; honest circuit-breaker-not-redactor framing matching ADR-1577 / security-model. - CONTEXT.md: add the 'Untrusted-input boundary' seam glossary entry. Verified: lint:docs ok; config-field-docs + contributor-standards green. * test(#1577): make read-injection property test git-text, not binary trek-e nit (and more): the file embedded a raw U+FFFF AND a raw NUL byte as degenerate-edge inputs. The NUL is what actually made git classify it binary (git binary = NUL in first 8K). Replace both with text-safe escapes that keep the identical runtime values: '\\x00' and String.fromCodePoint(0xFFFF). File now diffs/blames line-by-line. Verified: property test 2/2; no NUL/raw-noncharacter bytes remain. * docs(#1577): align untrusted boundary docs Name all 10 ingress agents in INVENTORY/security-model and allowlist the intentional read-injection property corpus for the prompt-injection scanner. * docs(#1577): align ADR ingest agent count Update ADR-1577 from 8 to 10 ingest agents so it matches the actual boundary include set and the rest of the docs. --------- Co-authored-by: Tom Boucher <trekkie@nomorestars.com>	2026-06-24 17:07:23 -04:00
Tom Boucher	207d8f1697	fix(#1626): make the security gate severity-aware via per-threat severity (#1635) ... workflow.security_block_on was documented as the minimum threat severity that blocks advancement, but threats carried no severity and the auditor's threats_open count (the SECURITY.md gate field) counted every open threat regardless of severity — so the threshold had no effect, and the auditor's block_on vocabulary (open/unregistered/none) did not even match the config enum (critical/high/medium/low/none). - planner: add a Severity column to the STRIDE threat register; assign severity per threat. - auditor: read severity; reconcile the <config> block_on domain to the severity enum; redefine threats_open as the count of OPEN threats whose severity is at or above block_on (none => 0). Below-threshold opens are reported as non-blocking and excluded from threats_open. - SECURITY.md template + planning-config.md reconciled. No gate-check site changed: threats_open == 0 stays the gate everywhere; only its computation is now severity-filtered. Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>	2026-06-23 19:04:13 -04:00
Tom Boucher	f9d9dfb4bc	fix(#1627): scale security rigor by ASVS level (planner disposition + auditor depth) (#1636) ... workflow.security_asvs_level was display-only — the planner hardcoded 'mitigate if ASVS L1 requires it' and the auditor only echoed the level, so L2/L3 behaved identically to L1. - New reference gsd-core/references/security-asvs-levels.md defines L1 (opportunistic), L2 (standard), L3 (comprehensive) for both planner threat disposition and auditor verification depth (higher = superset). - planner: disposition now scales with the configured ASVS level (no hardcoded L1) + @-pointer to the reference. - auditor: verification depth scales with asvs_level (L1 grep-presence, L2 boundary/vector check, L3 end-to-end trace + bypass check). - planning-config.md + INVENTORY updated; planner kept under its 48K cap by extracting the goal-backward worked example to planner-guidance.md. Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>	2026-06-23 18:48:58 -04:00
Dave	c09c13f295	enhance(verify-phase): node-test causation control — prove the RED is content-caused (#1346) ... The #1279 node-test machine-proof confirmed a known-bad subject drives the negative test RED, but could not distinguish a genuine content-violation from a deceptive test that reds merely because GSD_PROHIB_SUBJECT is set. Add an optional fifth flat scalar `check_clean_fixture` (-> CheckDescriptor.cleanFixture) threading a KNOWN-CLEAN control subject through projectProhibitions + descriptorFromProjection. When present, the prover also runs the check against the clean subject and requires GREEN, so fail-first is proven only when the check is RED on the violation AND GREEN on the clean subject (content-dependent). Opt-in and additive: absent a clean fixture the prover behaves exactly as post-#1314 (no control, documented residual), preserving the zero-authoring compose path; the lint-rule kind needs no analog (its subject IS the linted file, no env indirection). Coverage: RED-first deceptive case, positive, missing-clean fail-closed, round-trip read-back/emit, fast-check property extended to the 5th scalar, and an end-to-end COMPOSE capstone (honest vs deceptive). Docs: ADR-550 dated addendum, prohibition-probe reference, spec-phase + verify-phase workflows. Closes #1346 Claude-Session: https://claude.ai/code/session_01GsPRb8zvpcT7Eat6vZw8PX	2026-06-21 10:44:48 -04:00
Tom Boucher	b63a500fad	fix(#1505): extract context_guard step to reference file; fix allow-test-rule see ref ... - Extract execute-phase.md context_guard step prose to gsd-core/references/execute-phase-context-guard.md (@-ref lazy load), bringing execute-phase.md back under the ADR-857 phase-6 size ceiling (92914 < 93166 bytes) - Fix allow-test-rule comment: add `see #1452` per ADR-456 lint rule - Update feat-1452 tests to check the reference file for extracted content - Register execute-phase-context-guard.md in INVENTORY-MANIFEST.json and INVENTORY.md Workflow References section - Regenerate workflow-size-baseline.json after file shrinkage Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-06-20 18:38:20 -04:00
Tom Boucher	4eca5ac96c	feat(#1452): add workflow.context_guard_mode to guard execute-phase against context exhaustion ... Proactive checkpoint guard fires at each wave boundary before spawning agents. Self-assesses context pressure against context-budget.md degradation tiers and warns (warn, default) or auto-invokes /gsd:pause-work (auto) when POOR tier (70%+) is detected. Config key validated; defaults to \"warn\". Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-06-20 18:10:49 -04:00
Tom Boucher	c330f70f65	feat(#1494): add workflow.mvp_mode to VALID_CONFIG_KEYS; document code_review_command and plan_chunked in planning-config.md (#1500) ... * feat(#1494): add workflow.mvp_mode to VALID_CONFIG_KEYS; document code_review_command, plan_chunked, mvp_mode in planning-config.md Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * chore: backfill PR number 1500 in changeset Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> --------- Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-06-20 15:32:07 -04:00