Skip to content

v1.94.35.0 fix(contracts): run gstack-skill-contract through the host-resolved bin path (GST-1088 lane B follow-up) - #3957

Merged
capy-ai[bot] merged 17 commits into
mainfrom
capy/gst-1088-lane-b-bin-path
Nov 3, 2026
Merged

capy-ai[bot] merged 17 commits into
mainfrom
capy/gst-1088-lane-b-bin-path

Conversation

@garrytan

@garrytan garrytan commented Nov 3, 2026 •

Copy link
Copy Markdown
Owner
{
  "schema_version": 1,
  "pr": 3957,
  "head": "171c31cb9a201c430177a5d335221fdc33818747",
  "version": "1.94.29.0",
  "base": "main",
  "gate": {
    "targeted": {
      "pass": 1543,
      "fail": 0,
      "skip": 21
    }
  },
  "session_kind": "unattended",
  "head_tree": "9208517eabf7529b9813daeaf0030f61c910d5cb",
  "product_patch_id": "576b33f1db6ae8ce0c2bd77861b100495f4d3203",
  "gstack_version": "1.94.24.0",
  "review_depth": {
    "depth": "light",
    "class": "code",
    "runtime_lines": 10,
    "rules": "19881e1ecf7ed1ad",
    "base": "ade35bc81356e64c38772efa20b29021e69dcf51",
    "registry": "ade35bc81356e64c38772efa20b29021e69dcf51",
    "gate_scan": {
      "tagged": 0,
      "head": "171c31cb9a201c430177a5d335221fdc33818747"
    },
    "reasons": [
      "runtime=10"
    ],
    "risk": []
  },
  "pregate": "requires-remote cleared @9208517eabf7 (7 checks, current for 171c31cb9a20)",
  "requires_remote": [
    "free-tests.yml/macos-named-regressions: https://github.com/garrytan/gstack/actions/runs/41682383264/attempts/1",
    "windows-smoke.yml/windows-smoke: https://github.com/garrytan/gstack/actions/runs/41682383166/attempts/1",
    "pregate.json/remote_gate: remote-gate tree=9208517eabf7 tests=34985 skipped=112 runs=41682383264/1 profiles=ci-reuse/free-tests heads=171c31cb9a20 verified=2026-11-03T17:22:55.055Z"
  ],
  "issues": {
    "fixes": [],
    "refs": [
      3909
    ],
    "flakes": []
  },
  "docs": {
    "status": "current",
    "scope": "AGENTS.md, CLAUDE.md, CONTRIBUTING.md, README.md and docs/ against 1ccab0d8..3129fe82 (read-only)"
  },
  "review_loop": {
    "passes": 1,
    "blocking_found": 0,
    "deferred": 0,
    "noted": 1,
    "fixed": 0,
    "scope": "frozen",
    "rounds": 1,
    "ledgers_read": 1,
    "dispatches": {
      "specialist": 1
    }
  },
  "carry": {
    "from": "3129fe827dd1b0dcf6c299b7cf89a50b0be08b2a",
    "reason": "base merge + test-only change + release metadata",
    "product_patch_id": "576b33f1db6ae8ce0c2bd77861b100495f4d3203",
    "changed_paths": [
      "product:.gstack/pregate.json",
      "metadata:CHANGELOG.md",
      "product:CONTRIBUTING.md",
      "product:ETHOS.md",
      "metadata:VERSION",
      "metadata:agents-digest/gstack-AGENTS.md",
      "product:autoplan/sections/ceo-phase.md",
      "product:autoplan/sections/ceo-phase.md.tmpl",
      "product:bin/gstack-evidence",
      "product:bin/gstack-inventory",
      "product:bin/gstack-pregate",
      "product:bin/gstack-skill-bill",
      "product:capy/skill/gstack/factory.md",
      "product:docs/TESTING_INTERNALS.md",
      "product:docs/inventory.md",
      "product:docs/inventory/evidence.json",
      "product:docs/pregate.md",
      "product:docs/skills.md",
      "product:docs/standalone-binaries.md",
      "product:docs/troubleshooting.md",
      "product:eval-plan/SKILL.md",
      "product:eval-plan/SKILL.md.tmpl",
      "product:lib/ci-carry.ts",
      "product:lib/eval-plan.ts",
      "product:lib/inventory.ts",
      "product:lib/pregate/index.ts",
      "product:lib/result-codes-eval.ts",
      "product:lib/run-evidence.ts",
      "product:lib/ship-preflight.ts",
      "product:lib/ship-release.ts",
      "product:lib/skill-bill.ts",
      "metadata:package.json",
      "product:ship/DRIVER.md",
      "product:ship/DRIVER.md.tmpl",
      "test:test/capy-skill-wrapper.test.ts",
      "test:test/eval-plan-no-gstack-arm.test.ts",
      "test:test/eval-plan-results.test.ts",
      "test:test/evidence.test.ts",
      "test:test/fixtures/arm-benchmark/PREREG.md",
      "test:test/fixtures/inventory-grandfathered.json",
      "test:test/fixtures/model-panel/PREREG.md",
      "test:test/fixtures/module-size-ratchet.json",
      "test:test/fixtures/skill-md-size-allowlist.json",
      "test:test/gstack-clean-producers.test.ts",
      "test:test/helpers/free-fixtures-data.ts",
      "test:test/inventory.test.ts",
      "test:test/pregate-guards.test.ts",
      "test:test/ship-ci-carry.test.ts",
      "test:test/ship-preflight.test.ts",
      "test:test/ship-release.test.ts"
    ],
    "merge_commits": 1,
    "range_diff": "1:  b772dc620 =  1:  b772dc620 feat(skills): when-to-use descriptions, wider triggers and contract blocks for /investigate, /review, /qa, /verify (GST-1088 lane B)\n 2:  ebbe8a24b =  2:  ebbe8a24b fix(skills): keep core trigger phrases in /review and /qa descriptions; bare gstack-skill-contract name until lane A ships it\n 3:  7384ca0d4 =  3:  7384ca0d4 fix(skills): Done-means list leaves out steps a host suppresses and allows the small-diff skip (review fix round)\n 4:  5255dd9c2 =  4:  5255dd9c2 fix: pre-landing review fixes\n 5:  93d0a20dd =  5:  93d0a20dd fix: Done-means honors loadout section exclusions and the skill's own stop rules (Step 11 voices)\n 6:  3d16d8269 =  6:  3d16d8269 fix: the Done-means checker sentence follows an explicit CONTRACT_CHECKER_SHIPPED flag\n 7:  cc1539cd9 =  7:  cc1539cd9 test: cite GST-1088 gate rev 2 for the /review cap raises\n 8:  d2f1e2997 =  8:  d2f1e2997 Revert \"test: cite GST-1088 gate rev 2 for the /review cap raises\"\n 9:  cc9b9b9a2 =  9:  cc9b9b9a2 fix: trim /qa's opening sentence by 4 bytes so the branch fits main's qa parity cap\n10:  981a3ce3d = 10:  981a3ce3d fix: flip CONTRACT_CHECKER_SHIPPED now that bin/gstack-skill-contract ships, and re-pin /review by the measured 240 bytes\n11:  58aeee7c6 = 11:  58aeee7c6 docs: the changelog fragment says lane A's routing hook and contract check have landed\n12:  ed0c82ade = 12:  ed0c82ade fix(contracts): run gstack-skill-contract through the host-resolved bin path\n13:  3129fe827 = 13:  3129fe827 docs: changelog fragment wording for the bin-path fix",
    "covers": [
      "pre-landing review",
      "adversarial review",
      "gate integrity",
      "documentation audit"
    ],
    "reruns": [
      "CI at head (gstack-ship-queue ready requires it); test and helper code changed since the reviewed head runs there unreviewed"
    ],
    "ancestry": "linear",
    "trees": [
      "6b185296d7cebf82b89487f8e2c3666ea8103180"
    ]
  }
}

What changed

The Done means list now runs the checker through gstack's own bin folder instead of a bare command a standard install doesn't put on PATH

The "Done means" list in /investigate and /review now runs the contract checker through gstack's own bin folder, for example ~/.claude/skills/gstack/bin/gstack-skill-contract check on Claude Code, instead of a bare gstack-skill-contract.

Before/after as text: The Done means list now runs the checker through gstack's own bin folder instead of a bare command a standard install doesn't put on PATH
--- Before · review/SKILL.md § Done means at 1ccab0d · source text
+++ After · review/SKILL.md § Done means on this branch · source text
@@ ⋯ 4 unchanged ⋯ @@
 - `adversarial-pass`: an Agent dispatch for the adversarial pass, its prompt including "Think like an attacker and a chaos engineer".
 - `review-report`: a `Pre-Landing Review` with Gate edits filled in (`none: <why>` where the skill's own stop rule leaves one empty).
 
+`~/.claude/skills/gstack/bin/gstack-skill-contract check` reads them from the transcript (docs/skill-routing-and-contracts.md). For an item that truly does not apply, run `~/.claude/skills/gstack/bin/gstack-skill-contract waive <item> --reason "<why>"` (this session only; it counts as a miss).
-`gstack-skill-contract check` reads them from the transcript (docs/skill-routing-and-contracts.md). For an item that truly does not apply, run `gstack-skill-contract waive <item> --reason "<why>"` (this session only; it counts as a miss).

Why

A standard install doesn't put gstack's bin folder on PATH, so an agent that followed the list got "command not found" when it tried to record a waiver. The fix-check on #3909 at 58aeee7 found it, and it would skew GST-1089's contract adherence numbers if it shipped into the after pin. Every other gstack command in a skill already renders through the host-resolved bin path; this sentence now does too.

Blast radius

One sentence in two generated skills changes, on every host. Each host gets its own bin path from ctx.paths.binDir (~/.claude/skills/gstack/bin on Claude Code, $GSTACK_BIN on hosts that resolve it at run time). No other wording moves. The waiver parser (lib/skill-contract.ts:289) is unanchored, so a waive call through the full path still records the waiver.

Approvals

Merging main at 1ccab0d ran under GStack Minder's rule 19 grant (GSTA-36), relayed by the lane B parent. This is a bug fix with a failing test written first, so it took no plan approval (fix-lane exemption, R65).

APPROVALS_RENDER: redacted (repository visibility public); gstack-approvals list shows the full record

Summary

  • scripts/resolvers/contract-done-means.ts: renderDoneMeans(contract, binDir, checker) renders check and waive as binDir + '/gstack-skill-contract …'; generateContractDoneMeans passes ctx.paths.binDir.
  • test/skill-descriptions.test.ts: a new test checks every host in HOST_PATHS for /investigate and /review. It failed before the fix (111 pass, 1 fail: investigate on claude: ... must run ~/.claude/skills/gstack/bin/gstack-skill-contract check) and passes after.
  • Re-pins, by the measured delta only (no slack), because the sentence now carries the bin path:
    • test/helpers/carve-guards.ts /review maxSkeletonBytes: 60,014 → 60,070 (+56)
    • test/helpers/carve-guards.ts /review maxSizeRatio: 1.3283 → 1.3288 (union 144,203 / baseline 108,523 = 1.328778)
    • test/fixtures/skill-md-size-allowlist.json gstack-review: agy 104,728 → 104,752, codex 104,251 → 104,275, copilot 104,309 → 104,333, cursor 122,781 → 122,805, factory 126,554 → 126,578, gemini 104,131 → 104,155, kiro 122,605 → 122,629, opencode 123,403 → 123,427, slate 122,693 → 122,717 (+24 each)
    • /investigate: ratio 0.960 → 0.961, under its cap; no pin moves.
    • These pins still live in carve-guards.ts (lines 493 and 496) and the allowlist on 1ccab0d. v1.94.20.0 test(review): one file per review and harness pin, so independent repins merge cleanly (GST-1084) #3903's per-file pin files (test/fixtures/review-pins/, scripts/harness-pins/) hold none of them.
    • Fixed caps unchanged: 5315 ratchet, DRIVER.md 12,288 B, global skeleton 61,420, module 800, PART_MAX 24,576.

Test Coverage

Coverage: deterministic (each of 3 changed runtime files has a changed test naming it; no audit reader).

Pre-Landing Review

Pre-Landing Review: 5 issues — 0 auto-fixed, 0 asked, 0 deferred, 5 noted

REVIEW_LOOP: passes=1 blocking_found=0 deferred=0 noted=1 fixed=0 scope=frozen ledgers_read=1 dispatches=1 (plus one native adversarial pass, 4 findings, all noted)

Gate edits: 7 listed, 7 read, 0 findings — 0 kept, 0 restored, 0 open. The listed hunks are the re-pins above (test/helpers/carve-guards.ts, test/fixtures/skill-md-size-allowlist.json) and the new test.

Review notes

Below the blocking floor; recorded here and not fixed in this PR.

  • The Stop hook's block message (lib/skill-contract.ts:475) and docs/skill-routing-and-contracts.md:100 still name a bare gstack-skill-contract waive. Two reviewers found it. A dropped waiver already counts as a miss, so no test fails and no data is lost. It's a lane A follow-up.
  • The sentence's check names no --transcript/--skill flags, so running it as written exits 2. That wording came from v1.94.22.0 feat(skills): when-to-use descriptions, wider triggers and contract blocks for /investigate, /review, /qa, /verify (GST-1088 lane B) #3909, and this fix changes no other wording.
  • On hosts that use $GSTACK_BIN, an inline code span isn't run with the runtime prelude. Every other inline bin mention behaves the same, and these hosts have no contract hook.
  • The Claude path assumes a global install, like every other bin reference in Claude renders.

Issues

Refs #3909

Eval Results

bun run eval:select --base origin/main selects the dependents of scripts/resolvers/contract-done-means.ts; they run as this PR's ready --screen paid screen.

Documentation

Status: current. No documentation edits were needed or made (read-only mode).

Scope: base 1ccab0d..head 3129fe8 on capy/gst-1088-lane-b-bin-path. No staged, unstaged or untracked content. The 7 committed files are the changelog fragment, investigate/SKILL.md, review/SKILL.md (both generated), scripts/resolvers/contract-done-means.ts, test/fixtures/skill-md-size-allowlist.json, test/helpers/carve-guards.ts and test/skill-descriptions.test.ts. Discovery covered the candidate's authored doc inventory, with focused source verification of the changed contract.

Documentation health

  • changelog.d/capy-gst-1088-lane-b-bin-path.md: current. It accurately describes the resolver change and the tests, and uses plain language.
  • docs/skill-routing-and-contracts.md: current. Its bare gstack-skill-contract mentions are a command synopsis, not run instructions, so there is no contradiction. It was verified by grep, not a full read.
  • README.md, AGENTS.md, CLAUDE.md, CONTRIBUTING.md, ARCHITECTURE.md and the other docs: no mention of this change surface beyond the above, and no stale claims found. Not individually read.
  • CHANGELOG.md, TODOS.md and VERSION: out of scope for the ship-owned audit, untouched.

Coverage debt: the optional note on how to reach the checker (see decisions) is a gap only reported, not filled. There is no diagram drift.

Test plan

  • Failing test first: bun test test/skill-descriptions.test.ts failed 111/1 before the fix and passes after
  • Targeted lane at 3129fe8: 1543 pass, 0 fail, 21 skip
  • bun run typecheck: 0 errors; bun run gen:skill-docs --dry-run: 0 stale
  • Rule 25 trial merge with v1.94.29.0 R56: everything in gstack earns its place (principle, evidence gates, inventory) #3837 @9543745: allowlist conflict only, no overlapping keys, resolved as a union; typecheck 0 and size guard, parity, skill-descriptions, carve and context-budget tests 129 pass 0 fail on the merged tree
  • CI green at 3129fe8 (Free Tests run 41669942039)

garrytan and others added 14 commits November 3, 2026 02:14
…locks for /investigate, /review, /qa, /verify (GST-1088 lane B)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
…s; bare gstack-skill-contract name until lane A ships it

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
…llows the small-diff skip (review fix round)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
… stop rules (Step 11 voices)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
…CKER_SHIPPED flag

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
… qa parity cap

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
… ships, and re-pin /review by the measured 240 bytes

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
…check have landed

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
…in path

The Done-means check/waive sentence named a bare gstack-skill-contract,
which a standard install does not put on PATH (fix-check finding on #3909
at 58aeee7). Render it through ctx.paths.binDir like every other gstack bin,
with a failing-first test across every host. Re-pin /review by the measured
+56 skeleton bytes and +24 chars per non-Claude host render.

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
…-1088-lane-b-bin-path

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

# Conflicts:
#	investigate/SKILL.md
#	review/SKILL.md
#	scripts/resolvers/contract-done-means.ts
#	test/fixtures/skill-md-size-allowlist.json
#	test/helpers/carve-guards.ts
#	test/skill-descriptions.test.ts
@trunk-io

trunk-io Bot commented Nov 3, 2026 •

Copy link
Copy Markdown

😎 This pull request was merged.

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
@github-actions

github-actions Bot commented Nov 3, 2026

Copy link
Copy Markdown
Contributor

E2E Evals: ✅ PASS

[test:paid] VERDICT GREEN — lane gate/pr, attempt 1
  rule 12/12 · behavior 0/0 · judge 3/3 · quarantined 0 (0 failing the lane)
  SKIPPED 0 · INFRA 0 · INCOMPLETE 0 · unattributed 0 · ACTION REQUIRED 0
  wall 4m54s · cost $5.44 known + 1 of 15 trial(s) cost unknown

14 executed, 0 reused rule/judge records | 0 manual accepted (unscored; no score-cache credit) | $5.44 rule/judge cost | reconcile exit: 0

Profile: pr / mode pr; selected behaviors: 12, judges: 3. Deferred to scheduled/release coverage: 58 behaviors and 0 changed prompt files. Deferred checks did not run and receive no PR-pass credit.

Rule and judge shards
Shard Automated result Manual/unscored Executed Reused Status Cost
e2e-qa-callers/skill-e2e-qa-callers--review-exploratory-small-cli 1/1 0 1 0 ✅ $0.78
e2e-qa-callers/skill-e2e-qa-callers--ship-exploratory-plan-checks 1/1 0 1 0 ✅ $0.98
e2e/skill-e2e-investigate-owned-completion 1/1 0 1 0 ✅ $0.07
e2e-qa-callers/skill-e2e-qa-callers--ship-exploratory-late-input 1/1 0 1 0 ✅ $0.99
e2e-qa-callers/skill-e2e-qa-callers--ship-exploratory-small-cli 1/1 0 1 0 ✅ $1.01
llm-judge/skill-llm-eval 3/3 0 3 0 ✅ $0.18
e2e/skill-e2e-coverage-audit 1/1 0 1 0 ✅ $0.12
e2e/skill-e2e-investigate-owned-termination 2/2 0 2 0 ✅ $0.13
e2e/skill-e2e-test-value 1/1 0 1 0 ✅ $0.17
e2e-qa-callers/skill-e2e-qa-callers--ship-exploratory-unavailable 1/1 0 1 0 ✅ $0.78
e2e-review/skill-e2e-review 1/1 0 1 0 ✅ $0.23
Fail-closed reconciliation
[test:paid] report: 3/3 slices, 12 planned shards, tier=gate, attempt 1
[test:paid] coverage: profile=pr mode=pr; selected E2E=12, judges=3
[test:paid] deferred: 58 broad behaviors, 0 changed prompts without quick live coverage; these are not PR passes
  slice 1  passed             73s  test/skill-e2e-gate-integrity.test.ts
  slice 1  passed            102s  test/skill-llm-eval.test.ts
  slice 1  passed             23s  test/skill-e2e-coverage-audit.test.ts
  slice 2  passed            128s  test/skill-e2e-qa-callers.test.ts#ship-exploratory-late-input
  slice 2  passed            106s  test/skill-e2e-qa-callers.test.ts#review-exploratory-small-cli
  slice 2  passed            117s  test/skill-e2e-qa-callers.test.ts#ship-exploratory-plan-checks
  slice 2  passed            120s  test/skill-e2e-qa-callers.test.ts#ship-exploratory-small-cli
  slice 2  passed             24s  test/skill-e2e-investigate-owned-completion.test.ts
  slice 3  passed             43s  test/skill-e2e-test-value.test.ts
  slice 3  passed            100s  test/skill-e2e-qa-callers.test.ts#ship-exploratory-unavailable
  slice 3  passed             39s  test/skill-e2e-review.test.ts
  slice 3  passed             25s  test/skill-e2e-investigate-owned-termination.test.ts
[test:paid] collector final outcomes: 14 executed, 0 reused; 14 passed, 0 failed, 0 manual accepted (unscored; no score-cache credit) (14 attempt records from 11 collectors; every record counts)
[test:paid] deselected testcases, zero credit (46):
[test:paid]     25  outside the run's case selection
[test:paid]     20  sibling case of a case shard (its own shard runs it)
[test:paid]      1  periodic-tier case (this lane runs gate)
[test:paid] VERDICT GREEN — lane gate/pr, attempt 1
  rule 12/12 · behavior 0/0 · judge 3/3 · quarantined 0 (0 failing the lane)
  SKIPPED 0 · INFRA 0 · INCOMPLETE 0 · unattributed 0 · ACTION REQUIRED 0
  wall 4m54s · cost $5.44 known + 1 of 15 trial(s) cost unknown
[test:paid] exercised-path map (shadow, non-blocking): 4 MAP_GAP line(s) over 12 executed case(s) with recorded commands
  MAP_GAP ship-exploratory-late-input bin/gstack-review-plan: non-blocking for this PR; the next census refresh widens the map (docs/TESTING_INTERNALS.md#exercised-path-map)
  MAP_GAP ship-exploratory-plan-checks bin/gstack-review-plan: non-blocking for this PR; the next census refresh widens the map (docs/TESTING_INTERNALS.md#exercised-path-map)
  MAP_GAP ship-exploratory-small-cli bin/gstack-review-plan: non-blocking for this PR; the next census refresh widens the map (docs/TESTING_INTERNALS.md#exercised-path-map)
  MAP_GAP ship-exploratory-unavailable bin/gstack-review-plan: non-blocking for this PR; the next census refresh widens the map (docs/TESTING_INTERNALS.md#exercised-path-map)
[test:paid] report: every planned shard accounted and passed

Sliced lane: planner → duration-packed executors → fail-closed report. Behavior cases run a pre-registered 3-trial panel (PASS at 2/3 with no contract violation); a PASS 2/3 is shown with its failed trial, never as a clean pass. Reused results retain their original provenance and expiry.

Dispatched narrow screen (gstack-ship-queue ready --screen) of merge commit 7133f73cd14e13c69e061a928812a9a91ac5a6a6.

…-1088-lane-b-bin-path

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

# Conflicts:
#	test/fixtures/skill-md-size-allowlist.json
@github-actions github-actions Bot changed the title fix(contracts): run gstack-skill-contract through the host-resolved bin path (GST-1088 lane B follow-up) v1.94.35.0 fix(contracts): run gstack-skill-contract through the host-resolved bin path (GST-1088 lane B follow-up) Nov 3, 2026
@capy-ai
capy-ai Bot merged commit 8781d9c into main Nov 3, 2026
49 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant