Garry Tanandcapy-ai[bot] e9dfcf64f0 v1.91.137.0 feat(evals): gate-cost replay and merge-train cost model (GSTA-49 T0) (#3271) ...

* feat(evals): gate-cost replay and merge-train cost model (GSTA-49 T0)

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

* test(gate-cost): build fixtures in a temp dir instead of tracked test/fixtures

Tracked fixtures under test/fixtures need a FREE_FIXTURES entry, and
test/helpers/free-fixtures-data.ts is a paid-runner input that selects
every PR-profile case (57 cases + 24 judges, about 52 USD). The run tree
is now written to a temporary directory and the l3 screen costs are inline.

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

* chore: bump version and changelog (v1.91.137.0)

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>

2026-10-18 18:01:23 +00:00

..

app

feat: GStack Browser — double-click AI browser with anti-bot stealth (#695)

2026-04-04 10:17:05 -07:00

lib

v1.91.130.0 feat(pr-selection): exercised-path consumer map in shadow mode, MAP_GAP and census refresh (#3259)

2026-10-18 01:37:29 +00:00

resolvers

v1.91.127.0 fix(evals): paid-gate reliability round 2: docsync candidate protocol, scope-block detection, qa fix-loop budgets, lettered consent (#3260)

2026-10-18 00:37:43 +00:00

ubicloud

v1.91.132.0 feat(pregate): a remote gate for gstack's own pregate (Ubicloud or keyless local shards); concurrent next-version fetches (#3258)

2026-10-18 02:13:29 +00:00

analytics.ts

v1.91.11.0 refactor: one state-root rule, browse route table, shared shard engine, PTY harness split, MECE review resolvers (#3002)

2026-10-01 11:57:48 -07:00

archetypes.ts

feat: gstack v1 — simpler prompts + real LOC receipts (v1.0.0.0) (#1039)

2026-04-18 15:05:42 +08:00

brain-cache-spec.ts

v1.91.8.0 test: smaller suite, derived paid selection, retired never-green evals (#2994)

2026-09-29 08:58:49 -07:00

build-app.sh

v1.69.0.0 fix: the silent-failure wave — 6 fixes, 5 community PRs absorbed, tracker closed with receipts (#2666)

2026-08-24 08:55:58 -07:00

build-cso-windows.ps1

v1.91.19.0 fix: severe fix wave (truthful gates, no lost memory, every host finds gstack) (#3025)

2026-10-04 07:58:41 -07:00

build-cso.sh

v1.91.60.0 fix(windows): setup degrades instead of dying on a CSO failure, detects Smart App Control, Consolas terminal font (#3079)

2026-10-08 15:43:32 +00:00

build.sh

v1.91.65.0 fix(install): skills only for the agents you installed, every runtime-root file their skills read (#3080)

2026-10-08 16:17:18 +00:00

bump-harness-version.ts

v1.91.24.0 test: audit wave — PR evals keep finished work, honest ceilings, EVAL_POLICY v2, faster free suite (#3031)

2026-10-05 01:29:54 +00:00

capture-baseline.ts

v1.91.24.0 test: audit wave — PR evals keep finished work, honest ceilings, EVAL_POLICY v2, faster free suite (#3031)

2026-10-05 01:29:54 +00:00

compare-pr-version.ts

v1.91.122.0 fix(deps): clear the 2026-11-30 OSV suppressions, replace #3228, exempt dependabot from Version Gate rule 1 (#3239)

2026-10-17 16:47:50 +00:00

cso-attestation-evidence.ts

v1.87.0.0 feat: add verified CSO audits and replayable repair bundles (#2852)

2026-09-14 15:14:58 -07:00

cso-eval-output-redaction.ts

v1.91.75.0 fix(cso-eval): fail producer receipts only on credential-class output, naming the rules (#3120)

2026-10-11 01:18:11 +00:00

cso-eval-producer.ts

v1.91.135.0 fix(cso-eval): producer artifact inventory recognizes dependency-closure file names (#3265)

2026-10-18 09:12:41 +00:00

cso-eval-protocol.ts

v1.91.124.0 feat(cso): helper run clock (remainingSeconds), /cso time discipline, content-free producer tool timeline (#3242)

2026-10-17 17:19:52 +00:00

cso-eval.ts

v1.91.136.0 fix(cso-eval): falsely-tested count leaves out instrumentation claims; incidental claims certified by a passing signed-bundle replay (#3270)

2026-10-18 17:06:36 +00:00

cso-image-matrix.ts

v1.87.0.0 feat: add verified CSO audits and replayable repair bundles (#2852)

2026-09-14 15:14:58 -07:00

cso-native-evidence.ts

v1.91.25.0 feat(cso): private two-architecture release scoring + self-checking runtime release gates (#3036)

2026-10-04 21:47:21 -07:00

cso-public-ghcr.ts

v1.87.0.0 feat: add verified CSO audits and replayable repair bundles (#2852)

2026-09-14 15:14:58 -07:00

cso-requalification.ts

v1.91.25.0 feat(cso): private two-architecture release scoring + self-checking runtime release gates (#3036)

2026-10-04 21:47:21 -07:00

cso-runtime-promotion.ts

v1.91.84.0 fix(cso-eval): score real paid producer receipts — unique-path inventories, platform catalog equivalence, deterministic evaluation catalogs (#3127)

2026-10-11 15:17:32 +00:00

cso-sbom-packages-check.jq

v1.91.27.0 fix(cso): runtime staging can finish and a qualified scanner catalog can ship (#3041)

2026-10-05 10:00:47 -07:00

cso-sbom-packages.jq

v1.91.27.0 fix(cso): runtime staging can finish and a qualified scanner catalog can ship (#3041)

2026-10-05 10:00:47 -07:00

cso-scanner-catalog.ts

v1.87.0.0 feat: add verified CSO audits and replayable repair bundles (#2852)

2026-09-14 15:14:58 -07:00

cso-scanner-matrix.ts

v1.87.0.0 feat: add verified CSO audits and replayable repair bundles (#2852)

2026-09-14 15:14:58 -07:00

cso-staging-run.ts

v1.91.25.0 feat(cso): private two-architecture release scoring + self-checking runtime release gates (#3036)

2026-10-04 21:47:21 -07:00

cso-verify-runtime-base.ts

v1.87.0.0 feat: add verified CSO audits and replayable repair bundles (#2852)

2026-09-14 15:14:58 -07:00

declared-annotation.ts

v1.91.11.0 refactor: one state-root rule, browse route table, shared shard engine, PTY harness split, MECE review resolvers (#3002)

2026-10-01 11:57:48 -07:00

detect-bump.ts

v1.11.0.0 feat(ship): workspace-aware version allocation (#1168)

2026-04-23 23:03:27 -07:00

dev-skill.ts

v1.57.9.0 feat: source-clean gbrain render (dev-setup --out-dir + machine-wide gbrain-refresh) (#1951)

2026-06-09 22:29:23 -07:00

discover-skills.ts

v1.87.6.0 fix: make checks reliable and everyday validation faster (#2898)

2026-09-22 14:57:52 -04:00

e2e-shard-reuse.ts

v1.91.24.0 test: audit wave — PR evals keep finished work, honest ceilings, EVAL_POLICY v2, faster free suite (#3031)

2026-10-05 01:29:54 +00:00

eval-bg.ts

v1.91.24.0 test: audit wave — PR evals keep finished work, honest ceilings, EVAL_POLICY v2, faster free suite (#3031)

2026-10-05 01:29:54 +00:00

eval-compare.ts

v1.91.24.0 test: audit wave — PR evals keep finished work, honest ceilings, EVAL_POLICY v2, faster free suite (#3031)

2026-10-05 01:29:54 +00:00

eval-flake-rank.ts

v1.91.33.0 feat(ship): measurement bar with explained-red and pooled decisions, weekly flaky-test sweep on main (#3059)

2026-10-07 14:00:06 +00:00

eval-input-cache.ts

v1.91.12.0 v1.91.12.0: audit fix wave, ~11-minute paid eval lanes, eval reliability policy (#2999)

2026-10-01 13:55:16 -07:00

eval-list.ts

v1.91.24.0 test: audit wave — PR evals keep finished work, honest ceilings, EVAL_POLICY v2, faster free suite (#3031)

2026-10-05 01:29:54 +00:00

eval-readiness.ts

v1.91.134.0 feat(evals): readiness-gated paid dispatch (R1), verdict reuse (R2a/R2), receipt store that restores (#3262)

2026-10-18 06:03:18 +00:00

eval-select.ts

v1.91.24.0 test: audit wave — PR evals keep finished work, honest ceilings, EVAL_POLICY v2, faster free suite (#3031)

2026-10-05 01:29:54 +00:00

eval-summary.ts

v1.91.24.0 test: audit wave — PR evals keep finished work, honest ceilings, EVAL_POLICY v2, faster free suite (#3031)

2026-10-05 01:29:54 +00:00

eval-trial-series.ts

v1.91.24.0 test: audit wave — PR evals keep finished work, honest ceilings, EVAL_POLICY v2, faster free suite (#3031)

2026-10-05 01:29:54 +00:00

exercised-map.ts

v1.91.130.0 feat(pr-selection): exercised-path consumer map in shadow mode, MAP_GAP and census refresh (#3259)

2026-10-18 01:37:29 +00:00

external-skill-names.ts

v1.91.2.0 fix: consolidate gstack reliability wave (#2959)

2026-09-26 18:57:53 -04:00

free-test-durations-windows.json

v1.91.121.0 test(windows-seed): refresh the Windows duration seed from real windows-latest timings (#3243)

2026-10-17 06:44:32 +00:00

free-test-durations.json

v1.91.119.0 test(seed): refresh the free-suite duration seed (#3224) (#3238)

2026-10-17 05:14:19 +00:00

free-test-seed-allowlist.json

v1.91.119.0 test(seed): refresh the free-suite duration seed (#3224) (#3238)

2026-10-17 05:14:19 +00:00

garry-output-comparison.ts

fix: remove hardcoded author emails from throughput script

2026-04-18 15:36:50 +08:00

gate-cost-replay.ts

v1.91.137.0 feat(evals): gate-cost replay and merge-train cost model (GSTA-49 T0) (#3271)

2026-10-18 18:01:23 +00:00

gate-diff-calibrate.ts

feat: honest work and gate integrity — gate-edit detector, review/ship gate pass, zero-run green, claims-need-evidence kernel (#3114)

2026-10-10 22:33:01 +00:00

gate-reuse.ts

v1.91.134.0 feat(evals): readiness-gated paid dispatch (R1), verdict reuse (R2a/R2), receipt store that restores (#3262)

2026-10-18 06:03:18 +00:00

gen-agents-digest.ts

v1.75.0.0 feat: ponytail import wave — simplification review lens, arm benchmark, reuse ladder, instruction-tier digest (#2722)

2026-08-29 10:10:35 -07:00

gen-doctor-components.ts

v1.91.71.0 feat(install): gstack-doctor --check one-line verdict, gstack-capy-install, lazy Chromium (multi-agent wave PR A) (#3116)

2026-10-11 00:00:20 +00:00

gen-llms-txt.ts

v1.91.66.0 fix: /autoplan without permission prompts after background reviewers, 34 contributor PRs from the backlog triage (#3087)

2026-10-08 17:40:06 +00:00

gen-ship-goldens.ts

v1.91.77.0 feat(ship): two-stage gstack-pregate, gstack-regen generated-file registry, intent-based issue links (multi-agent wave C-pregate) (#3121)

2026-10-11 01:57:16 +00:00

gen-skill-docs.ts

v1.91.97.0 fix: closable unattended /autoplan runs on Capy (field wave B, items 6-13) (#3165)

2026-10-13 15:17:39 +00:00

gstack-schema-pack.ts

v1.52.1.0 feat: brain-aware planning — 5 skills read structured gbrain context before asking (#1742)

2026-05-29 08:35:00 -07:00

harness-version.json

v1.91.130.0 feat(pr-selection): exercised-path consumer map in shadow mode, MAP_GAP and census refresh (#3259)

2026-10-18 01:37:29 +00:00

host-config-export.ts

v1.64.1.0 v1.64.1.0: the code-smell fix wave — every pipeline guard now provably fires (net −24,943 lines) (#2572)

2026-08-15 09:37:04 -07:00

host-config.ts

v1.91.16.0 fix: safe installs and upgrades, skills start everywhere, truthful completion (parallel wave) (#3014)

2026-10-03 07:54:09 -07:00

jargon-list.json

feat: gstack v1 — simpler prompts + real LOC receipts (v1.0.0.0) (#1039)

2026-04-18 15:05:42 +08:00

judge-calibration-configs.ts

v1.91.30.0 fix: honest eval failure reports, Bun floor with safe auto-update, make-pdf and /review on Codex (#3033)

2026-10-06 18:07:50 +00:00

judge-calibration.ts

v1.91.18.0 feat: Opus 5.5 cleanup follow-ups — scoped transcript consent, opt-in Claude overlay, safety-rule evals, calibrated judges (#3023)

2026-10-03 18:49:34 -07:00

judge-controls.ts

v1.91.30.0 fix: honest eval failure reports, Bun floor with safe auto-update, make-pdf and /review on Codex (#3033)

2026-10-06 18:07:50 +00:00

measure-journal-read.ts

v1.91.34.0 fix: /autoplan on current Claude Code and long sessions, quieter pre-push and sync guards, gstack-post, gstack-doctor (#3069)

2026-10-07 23:29:05 +00:00

model-policy-freshness.ts

v1.91.45.0 feat: configurable model tiers and advisory freshness (#3073)

2026-10-08 08:00:08 +00:00

models.ts

v1.91.18.0 feat: Opus 5.5 cleanup follow-ups — scoped transcript consent, opt-in Claude overlay, safety-rule evals, calibrated judges (#3023)

2026-10-03 18:49:34 -07:00

one-way-doors.ts

v1.61.0.0 fix wave: guards failing open / silent failures (9 fixes, 4 community PRs absorbed) (#2472)

2026-08-08 09:28:45 -07:00

paid-test-durations.json

v1.91.30.0 fix: honest eval failure reports, Bun floor with safe auto-update, make-pdf and /review on Codex (#3033)

2026-10-06 18:07:50 +00:00

pr-dependencies.ts

v1.91.130.0 feat(pr-selection): exercised-path consumer map in shadow mode, MAP_GAP and census refresh (#3259)

2026-10-18 01:37:29 +00:00

preflight-agent-sdk.ts

v1.91.24.0 test: audit wave — PR evals keep finished work, honest ceilings, EVAL_POLICY v2, faster free suite (#3031)

2026-10-05 01:29:54 +00:00

preflight-codex-overlap.ts

v1.91.45.0 feat: configurable model tiers and advisory freshness (#3073)

2026-10-08 08:00:08 +00:00

prompt-audit-manifest.ts

v1.91.18.0 feat: Opus 5.5 cleanup follow-ups — scoped transcript consent, opt-in Claude overlay, safety-rule evals, calibrated judges (#3023)

2026-10-03 18:49:34 -07:00

psychographic-signals.ts

v1.91.12.0 v1.91.12.0: audit fix wave, ~11-minute paid eval lanes, eval reliability policy (#2999)

2026-10-01 13:55:16 -07:00

question-registry.ts

feat: honest work and gate integrity — gate-edit detector, review/ship gate pass, zero-run green, claims-need-evidence kernel (#3114)

2026-10-10 22:33:01 +00:00

recover-receipts.ts

v1.91.24.0 test: audit wave — PR evals keep finished work, honest ceilings, EVAL_POLICY v2, faster free suite (#3031)

2026-10-05 01:29:54 +00:00

replay-pr-selection.ts

v1.91.129.0 fix(pr-selection): scan shebang and .ts imports, fail closed on scan errors, report cost per run (#3257)

2026-10-18 01:22:15 +00:00

resolve-codex-generation-model.ts

v1.91.45.0 feat: configurable model tiers and advisory freshness (#3073)

2026-10-08 08:00:08 +00:00

retired-command.ts

v1.91.24.0 test: audit wave — PR evals keep finished work, honest ceilings, EVAL_POLICY v2, faster free suite (#3031)

2026-10-05 01:29:54 +00:00

sandbox-doctor.sh

v1.75.0.0 feat: ponytail import wave — simplification review lens, arm benchmark, reuse ladder, instruction-tier digest (#2722)

2026-08-29 10:10:35 -07:00

setup-scc.sh

feat: gstack v1 — simpler prompts + real LOC receipts (v1.0.0.0) (#1039)

2026-04-18 15:05:42 +08:00

ship-measure-sweep.ts

v1.91.33.0 feat(ship): measurement bar with explained-red and pooled decisions, weekly flaky-test sweep on main (#3059)

2026-10-07 14:00:06 +00:00

ship-measure.ts

v1.91.34.0 fix: /autoplan on current Claude Code and long sessions, quieter pre-push and sync guards, gstack-post, gstack-doctor (#3069)

2026-10-07 23:29:05 +00:00

skill-check.ts

v1.87.6.0 fix: make checks reliable and everyday validation faster (#2898)

2026-09-22 14:57:52 -04:00

slop-diff.ts

security: tunnel dual-listener + SSRF + envelope + path wave (v1.6.0.0) (#1137)

2026-04-21 21:58:27 -07:00

test-free-shards.ts

v1.91.132.0 feat(pregate): a remote gate for gstack's own pregate (Ubicloud or keyless local shards); concurrent next-version fetches (#3258)

2026-10-18 02:13:29 +00:00

test-health-report.ts

v1.91.134.0 feat(evals): readiness-gated paid dispatch (R1), verdict reuse (R2a/R2), receipt store that restores (#3262)

2026-10-18 06:03:18 +00:00

test-paid-shards.ts

v1.91.30.0 fix: honest eval failure reports, Bun floor with safe auto-update, make-pdf and /review on Codex (#3033)

2026-10-06 18:07:50 +00:00

test-pr-profile.ts

v1.91.130.0 feat(pr-selection): exercised-path consumer map in shadow mode, MAP_GAP and census refresh (#3259)

2026-10-18 01:37:29 +00:00

test-strict-output.ts

v1.91.11.0 refactor: one state-root rule, browse route table, shared shard engine, PTY harness split, MECE review resolvers (#3002)

2026-10-01 11:57:48 -07:00

typecheck-test-baseline.json

v1.91.122.0 fix(deps): clear the 2026-11-30 OSV suppressions, replace #3228, exempt dependabot from Version Gate rule 1 (#3239)

2026-10-17 16:47:50 +00:00

typecheck-test.ts

v1.91.19.0 fix: severe fix wave (truthful gates, no lost memory, every host finds gstack) (#3025)

2026-10-04 07:58:41 -07:00

write-version-files.sh

v1.42.0.0 Daegu wave: 23 community-filed bugs + PTY classifier enforcement (24 bisect commits) (#1594)

2026-05-20 07:35:01 -07:00