[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7C51CCF329M80BCRH2E7B2R":3,"skill-same:01M7C51CCF329M80BCRH2E7B2R":212},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":25,"commitSha":26,"skillDirectory":27,"forksCount":14,"repositoryHtmlUrl":28,"repositoryHomepageUrl":6,"ownerHtmlUrl":29,"ownerBlogUrl":30,"ownerDescription":6,"updatedAt":31,"install":32,"resources":44},"01M7C51CCF329M80BCRH2E7B2R","qa-verdict-gate",null,"Kill-gate: QA PASS\u002FFAIL\u002FINCONCLUSIVE + qa-report.md. Triggers: QA verdict, QA done\u002Fcomplete, pre-merge QA, false-FAIL.","EtanHey\u002Fgolems","EtanHey","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F93338119?v=4","01M7C4SKC9HJC3B9TAX0D9MZ1E","Etan Heyman","01M7C4SKCT59Z49TTN4WWDDEYD",0,"Apache-2.0","https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: qa-verdict-gate\ndescription: \"Kill-gate: QA PASS\u002FFAIL\u002FINCONCLUSIVE + qa-report.md. Triggers: QA verdict, QA done\u002Fcomplete, pre-merge QA, false-FAIL.\"\ndisable-model-invocation: true\n---\n\n# Skill: QA Verdict-Integrity Kill-Gate (gen-18 Track 2 #6)\n\n> Fleet law: canon #4 owns user-visible completion; canon #3 owns evidence. This skill enforces QA verdict integrity, report artifacts, and rendered-state evidence.\n\n## Scope\n\nFAIL is only for confirmed-observed failures; INCONCLUSIVE is for when the path couldn't be reached. A qa-report.md artifact must exist.\n\n## What It Is\n\nA deterministic detector over a QA transcript\u002Fverdict: a QA result is a **verdict-integrity violation**\nunless the agent's stated verdict matches the OBSERVED evidence in the same turn AND a terminal\n`qa-report.md` artifact exists. This is the MECHANICAL gate that `\u002Fqa-video` and `\u002Fnever-fabricate`\ndescribe in prose. The pinned RED\u002FGREEN transcript fixtures ARE the replayable gate (R-003\u002FR-014 pattern,\nT6 deterministic smoke-spec shape). Sibling of `\u002Ffalse-green-gate` — same `lib\u002Ftranscript.mjs`, same\n`buildEvidence` discipline (evidence comes from tool calls + tool_result OUTPUTS, never assistant prose).\n\n## The Rule — tri-state, evidence-anchored\n\n| Detector code | Fires when | The honest verdict instead |\n|---|---|---|\n| `QA_FAIL_WITHOUT_OBSERVATION` | a **FAIL** with no same-turn observed-failure evidence (screenshot, click that reached the surface, observed error, or matching qa-video inspected-ledger rows in `tool_result` outputs) | `INCONCLUSIVE` |\n| `QA_UNREACHED_NOT_INCONCLUSIVE` | the transcript shows the path **couldn't be reached** (\"couldn't load\", \"element not found\", \"blocked at step 0\", `net::ERR_*`), but the verdict is **FAIL or PASS** | `INCONCLUSIVE` |\n| `QA_NO_REPORT_ARTIFACT` | a QA **\"done\u002Fcomplete\"** claim with no `qa-report.md` written this turn carrying the checklist items (≥2) | write the report first |\n| `QA_DECISION_CLAIM_NO_VISUAL_EVIDENCE` | a terminal QA\u002Fcompletion answer that feeds an Etan merge\u002Fship\u002FQA decision has no screenshot\u002Fdashboard\u002Frendered-state reference embedded in that same answer | embed the visual evidence reference where Etan answers |\n| `QA_USER_VISIBLE_WITHOUT_RENDERED_EVIDENCE` | a **PASS\u002Fready\u002Fdone** verdict about a user-facing surface is backed only by merge\u002Fplumbing state, with no same-turn screenshot\u002Fclick\u002Frendered page\u002FHTTP 2xx probe | probe the rendered user-visible surface first |\n\nPASS = the stated verdict matches the observed evidence **and** the `qa-report.md` artifact exists.\n\"Same turn\" = the events since the last human message — the probe tool calls **and their results** the\nagent ran before its verdict. Negated\u002Fin-progress statements (\"not done yet\") are not verdicts. An explicit\n`Verdict: INCONCLUSIVE` line is authoritative over loose pass\u002Ffail words elsewhere in the prose.\n\n### Recorded-frame observation\n\nFor qa-video, same-turn tool outputs may supply headerless inspected-ledger rows:\n`pts\u003CTAB>absolute frame path\u003CTAB>reader: why\u003CTAB>finding-id[\u003CTAB>sha256]`.\nLegacy four-column rows are accepted. Five-column rows require a 64-hex-digit\nSHA-256 field; other row shapes are rejected. Read the run's\n`inspected.tsv` (or `logs\u002Finspected*.tsv`) and include its rows in the tool output.\nPTS must be finite and nonnegative; the canonical absolute `.png`\u002F`.jpg`\u002F`.jpeg`\npath must stay under the run directory of a same-turn Write\u002FEdit\u002FMultiEdit to\n`\u002F…\u002Fqa-video\u002F\u003Crun>\u002Fqa-report.md` (namespaced tool names are supported).\nThere must be exactly one such run root. Every named FAIL in assistant text or\nthe report needs a matching row: `F1 FAIL` or `### F1: title — **FAIL**`, with\ncase-sensitive alphanumeric\u002Fhyphen IDs containing a digit (e.g. `S1-F6r`).\nRows for another finding\u002Frun, earlier-turn rows, assistant prose, and CONFIRMED\nrecords or verification summaries alone do not count. This checks inspection\nreceipts, not image contents, before\u002Fafter sufficiency, or verifier correctness.\nOnly the FAIL observation check accepts these receipts; the other four checks\nremain unchanged, including the requirement to write the report this turn.\n\n## Why FAIL is reserved\n\nA false-FAIL (codex `019ee5ea#1`) and a truncated QA with no artifact (`019ee493#2`\u002F`#4`) are the two\nspecimens this closes. A FAIL emitted when the regression path was never reached is *worse than no QA* —\nit sends a worker chasing a phantom bug. FAIL must come from a confirmed observation; everything the run\ncould not actually exercise is `INCONCLUSIVE`, surfaced honestly.\n\n## How \u002Fpr-loop + \u002Fqa-video Consume It\n\nBefore a QA seat emits a verdict or a \"QA complete\" message, run the gate on the turn — `\u002Fqa-verdict-gate`\n(`bun skills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fscripts\u002Fqa-verdict-gate-cli.mjs \u003Ctranscript|->`, exit 3 = FLAG).\nA FLAG means the verdict is not yet earned: reach the surface and observe, or downgrade FAIL→INCONCLUSIVE,\nwrite the missing `qa-report.md`, embed the visual reference in decision-feeding answers, or probe the\nrendered user-visible surface. Compose with `\u002Ffalse-green-gate` (live-outcome probe for completion claims)\nand `\u002Fnever-fabricate` (read the output before reporting).\n\n## Run It\n\n```bash\nbun test skills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fevals\u002Fqa-verdict-gate.test.mjs   # replay (CI-safe)\npython3 skills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fevals\u002Frun_suite.py                # hook schema + fail-open\nbun skills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fscripts\u002Fqa-verdict-gate-cli.mjs \u003Ctranscript.jsonl|->\n```\n\nProgrammatic: `import { detectQaVerdict } from \".\u002Fsrc\u002Fqa-verdict-gate.mjs\"` → `{ verdict, claim, qaVerdict, violations }`.\n\n## Stated Limits (honesty rule)\n\n- Evidence is marker-anchored over the same-turn blob; a probe described but not actually run can read as\n  present. The fixtures pin the known specimens; new evasion shapes are added as RED fixtures (R-003 model).\n- \"Same turn\" is bounded by the last human message; a probe run two turns earlier does not count (by design).\n- Observed-error \u002F screenshot tokens in a `tool_result` only count when a Bash command or a browser tool\n  actually drove a surface this turn — a passive `Read` of an old log cannot supply them (same-turn-scope).\n  The recorded-frame route instead accepts a same-turn ledger Read under the contract above.\n- Decision-feeding evidence is checked on the terminal answer text: a screenshot mentioned only earlier in\n  the turn does not help Etan answer the decision prompt.\n\n## Provenance\n\nRED specimens: codex\u002F`019ee5ea#1` (false-FAIL, path unreachable), codex\u002F`019ee493#2`\u002F`#4` (truncated QA,\nno artifact), a FAIL with zero observed-failure evidence. Evasion REDs: prose-bypass (narrative \"observed\nerror\"), passive-read-supplies-error (same-turn-scope), false-PASS over an unreachable surface,\nreport-stub (artifact with no checklist), prose-unreached-still-FAIL (negation), namespaced-MCP-Write\n(tool-name normalization), navigate-only-clears-FAIL (cursor-bugbot L195 — bare `browser_navigate` does\nnot exercise the failing flow), decision-feeding bare verdict with no visual evidence reference,\nuser-facing dashboard ready claim backed only by merge state. GREEN references: FAIL with an observed-error\nscreenshot + report, INCONCLUSIVE correctly used when unreachable, PASS with a full report, in-progress\n(N\u002FA), verdict-pending (cursor-bugbot L260 — \"step 1 done, verdict still pending\" is not a settled verdict),\ndecision-feeding claim with the screenshot\u002Fdashboard reference embedded in the same terminal answer.\n","# Skill: QA Verdict-Integrity Kill-Gate (gen-18 Track 2 #6)\n\n> Fleet law: canon #4 owns user-visible completion; canon #3 owns evidence. This skill enforces QA verdict integrity, report artifacts, and rendered-state evidence.\n\n## Scope\n\nFAIL is only for confirmed-observed failures; INCONCLUSIVE is for when the path couldn't be reached. A qa-report.md artifact must exist.\n\n## What It Is\n\nA deterministic detector over a QA transcript\u002Fverdict: a QA result is a **verdict-integrity violation**\nunless the agent's stated verdict matches the OBSERVED evidence in the same turn AND a terminal\n`qa-report.md` artifact exists. This is the MECHANICAL gate that `\u002Fqa-video` and `\u002Fnever-fabricate`\ndescribe in prose. The pinned RED\u002FGREEN transcript fixtures ARE the replayable gate (R-003\u002FR-014 pattern,\nT6 deterministic smoke-spec shape). Sibling of `\u002Ffalse-green-gate` — same `lib\u002Ftranscript.mjs`, same\n`buildEvidence` discipline (evidence comes from tool calls + tool_result OUTPUTS, never assistant prose).\n\n## The Rule — tri-state, evidence-anchored\n\n| Detector code | Fires when | The honest verdict instead |\n|---|---|---|\n| `QA_FAIL_WITHOUT_OBSERVATION` | a **FAIL** with no same-turn observed-failure evidence (screenshot, click that reached the surface, observed error, or matching qa-video inspected-ledger rows in `tool_result` outputs) | `INCONCLUSIVE` |\n| `QA_UNREACHED_NOT_INCONCLUSIVE` | the transcript shows the path **couldn't be reached** (\"couldn't load\", \"element not found\", \"blocked at step 0\", `net::ERR_*`), but the verdict is **FAIL or PASS** | `INCONCLUSIVE` |\n| `QA_NO_REPORT_ARTIFACT` | a QA **\"done\u002Fcomplete\"** claim with no `qa-report.md` written this turn carrying the checklist items (≥2) | write the report first |\n| `QA_DECISION_CLAIM_NO_VISUAL_EVIDENCE` | a terminal QA\u002Fcompletion answer that feeds an Etan merge\u002Fship\u002FQA decision has no screenshot\u002Fdashboard\u002Frendered-state reference embedded in that same answer | embed the visual evidence reference where Etan answers |\n| `QA_USER_VISIBLE_WITHOUT_RENDERED_EVIDENCE` | a **PASS\u002Fready\u002Fdone** verdict about a user-facing surface is backed only by merge\u002Fplumbing state, with no same-turn screenshot\u002Fclick\u002Frendered page\u002FHTTP 2xx probe | probe the rendered user-visible surface first |\n\nPASS = the stated verdict matches the observed evidence **and** the `qa-report.md` artifact exists.\n\"Same turn\" = the events since the last human message — the probe tool calls **and their results** the\nagent ran before its verdict. Negated\u002Fin-progress statements (\"not done yet\") are not verdicts. An explicit\n`Verdict: INCONCLUSIVE` line is authoritative over loose pass\u002Ffail words elsewhere in the prose.\n\n### Recorded-frame observation\n\nFor qa-video, same-turn tool outputs may supply headerless inspected-ledger rows:\n`pts\u003CTAB>absolute frame path\u003CTAB>reader: why\u003CTAB>finding-id[\u003CTAB>sha256]`.\nLegacy four-column rows are accepted. Five-column rows require a 64-hex-digit\nSHA-256 field; other row shapes are rejected. Read the run's\n`inspected.tsv` (or `logs\u002Finspected*.tsv`) and include its rows in the tool output.\nPTS must be finite and nonnegative; the canonical absolute `.png`\u002F`.jpg`\u002F`.jpeg`\npath must stay under the run directory of a same-turn Write\u002FEdit\u002FMultiEdit to\n`\u002F…\u002Fqa-video\u002F\u003Crun>\u002Fqa-report.md` (namespaced tool names are supported).\nThere must be exactly one such run root. Every named FAIL in assistant text or\nthe report needs a matching row: `F1 FAIL` or `### F1: title — **FAIL**`, with\ncase-sensitive alphanumeric\u002Fhyphen IDs containing a digit (e.g. `S1-F6r`).\nRows for another finding\u002Frun, earlier-turn rows, assistant prose, and CONFIRMED\nrecords or verification summaries alone do not count. This checks inspection\nreceipts, not image contents, before\u002Fafter sufficiency, or verifier correctness.\nOnly the FAIL observation check accepts these receipts; the other four checks\nremain unchanged, including the requirement to write the report this turn.\n\n## Why FAIL is reserved\n\nA false-FAIL (codex `019ee5ea#1`) and a truncated QA with no artifact (`019ee493#2`\u002F`#4`) are the two\nspecimens this closes. A FAIL emitted when the regression path was never reached is *worse than no QA* —\nit sends a worker chasing a phantom bug. FAIL must come from a confirmed observation; everything the run\ncould not actually exercise is `INCONCLUSIVE`, surfaced honestly.\n\n## How \u002Fpr-loop + \u002Fqa-video Consume It\n\nBefore a QA seat emits a verdict or a \"QA complete\" message, run the gate on the turn — `\u002Fqa-verdict-gate`\n(`bun skills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fscripts\u002Fqa-verdict-gate-cli.mjs \u003Ctranscript|->`, exit 3 = FLAG).\nA FLAG means the verdict is not yet earned: reach the surface and observe, or downgrade FAIL→INCONCLUSIVE,\nwrite the missing `qa-report.md`, embed the visual reference in decision-feeding answers, or probe the\nrendered user-visible surface. Compose with `\u002Ffalse-green-gate` (live-outcome probe for completion claims)\nand `\u002Fnever-fabricate` (read the output before reporting).\n\n## Run It\n\n```bash\nbun test skills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fevals\u002Fqa-verdict-gate.test.mjs   # replay (CI-safe)\npython3 skills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fevals\u002Frun_suite.py                # hook schema + fail-open\nbun skills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fscripts\u002Fqa-verdict-gate-cli.mjs \u003Ctranscript.jsonl|->\n```\n\nProgrammatic: `import { detectQaVerdict } from \".\u002Fsrc\u002Fqa-verdict-gate.mjs\"` → `{ verdict, claim, qaVerdict, violations }`.\n\n## Stated Limits (honesty rule)\n\n- Evidence is marker-anchored over the same-turn blob; a probe described but not actually run can read as\n  present. The fixtures pin the known specimens; new evasion shapes are added as RED fixtures (R-003 model).\n- \"Same turn\" is bounded by the last human message; a probe run two turns earlier does not count (by design).\n- Observed-error \u002F screenshot tokens in a `tool_result` only count when a Bash command or a browser tool\n  actually drove a surface this turn — a passive `Read` of an old log cannot supply them (same-turn-scope).\n  The recorded-frame route instead accepts a same-turn ledger Read under the contract above.\n- Decision-feeding evidence is checked on the terminal answer text: a screenshot mentioned only earlier in\n  the turn does not help Etan answer the decision prompt.\n\n## Provenance\n\nRED specimens: codex\u002F`019ee5ea#1` (false-FAIL, path unreachable), codex\u002F`019ee493#2`\u002F`#4` (truncated QA,\nno artifact), a FAIL with zero observed-failure evidence. Evasion REDs: prose-bypass (narrative \"observed\nerror\"), passive-read-supplies-error (same-turn-scope), false-PASS over an unreachable surface,\nreport-stub (artifact with no checklist), prose-unreached-still-FAIL (negation), namespaced-MCP-Write\n(tool-name normalization), navigate-only-clears-FAIL (cursor-bugbot L195 — bare `browser_navigate` does\nnot exercise the failing flow), decision-feeding bare verdict with no visual evidence reference,\nuser-facing dashboard ready claim backed only by merge state. GREEN references: FAIL with an observed-error\nscreenshot + report, INCONCLUSIVE correctly used when unreachable, PASS with a full report, in-progress\n(N\u002FA), verdict-pending (cursor-bugbot L260 — \"step 1 done, verdict still pending\" is not a settled verdict),\ndecision-feeding claim with the screenshot\u002Fdashboard reference embedded in the same terminal answer.\n",{"name":5,"description":7,"disable-model-invocation":24},true,"master","3400bc971628f00ab7bf1063993477e211a5508a","skills\u002Fgolem-powers\u002Fqa-verdict-gate","https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems","https:\u002F\u002Fgithub.com\u002FEtanHey","etanheyman.com","2026-11-15 03:05:08",{"source":8,"skillName":5,"commands":33,"prompt":43,"zipAvailable":24},[34,37,40],{"tool":35,"command":36},"npx","npx skills add EtanHey\u002Fgolems --skill qa-verdict-gate",{"tool":38,"command":39},"bunx","bunx skills add EtanHey\u002Fgolems --skill qa-verdict-gate",{"tool":41,"command":42},"pnpm","pnpm dlx skills add EtanHey\u002Fgolems --skill qa-verdict-gate","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002FSKILL.md",[45,51,55,59,63,67,71,75,79,83,87,91,95,99,103,107,111,115,119,123,127,131,135,139,143,147,151,155,159,162,166,170,174,178,182,188,192,196,200,204,208],{"path":46,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":49,"sourceHtmlUrl":50},"evals\u002Ffixtures\u002Fgreen\u002F01-fail-observed-error-with-report.json","REFERENCE","参考文档",1930,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fevals\u002Ffixtures\u002Fgreen\u002F01-fail-observed-error-with-report.json",{"path":52,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":53,"sourceHtmlUrl":54},"evals\u002Ffixtures\u002Fgreen\u002F02-inconclusive-unreachable.json",1531,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fevals\u002Ffixtures\u002Fgreen\u002F02-inconclusive-unreachable.json",{"path":56,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":57,"sourceHtmlUrl":58},"evals\u002Ffixtures\u002Fgreen\u002F03-pass-full-report.json",1869,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fevals\u002Ffixtures\u002Fgreen\u002F03-pass-full-report.json",{"path":60,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":61,"sourceHtmlUrl":62},"evals\u002Ffixtures\u002Fgreen\u002F04-in-progress-na.json",460,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fevals\u002Ffixtures\u002Fgreen\u002F04-in-progress-na.json",{"path":64,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":65,"sourceHtmlUrl":66},"evals\u002Ffixtures\u002Fgreen\u002F05-verdict-pending-na.json",531,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fevals\u002Ffixtures\u002Fgreen\u002F05-verdict-pending-na.json",{"path":68,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":69,"sourceHtmlUrl":70},"evals\u002Ffixtures\u002Fgreen\u002F06-decision-claim-with-visual-evidence.json",1609,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fevals\u002Ffixtures\u002Fgreen\u002F06-decision-claim-with-visual-evidence.json",{"path":72,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":73,"sourceHtmlUrl":74},"evals\u002Ffixtures\u002Fgreen\u002F07-generic-decision-word-not-etan-decision.json",1326,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fevals\u002Ffixtures\u002Fgreen\u002F07-generic-decision-word-not-etan-decision.json",{"path":76,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":77,"sourceHtmlUrl":78},"evals\u002Ffixtures\u002Fgreen\u002F08-quoted-verdict-word-not-a-qa-claim.json",810,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fevals\u002Ffixtures\u002Fgreen\u002F08-quoted-verdict-word-not-a-qa-claim.json",{"path":80,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":81,"sourceHtmlUrl":82},"evals\u002Ffixtures\u002Fgreen\u002F09-prose-blocked-and-pass-no-qa-context.json",1150,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fevals\u002Ffixtures\u002Fgreen\u002F09-prose-blocked-and-pass-no-qa-context.json",{"path":84,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":85,"sourceHtmlUrl":86},"evals\u002Ffixtures\u002Fgreen\u002F10-video-inspected-frames.json",1856,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fevals\u002Ffixtures\u002Fgreen\u002F10-video-inspected-frames.json",{"path":88,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":89,"sourceHtmlUrl":90},"evals\u002Ffixtures\u002Fgreen\u002F11-video-ledger-only.json",1685,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fevals\u002Ffixtures\u002Fgreen\u002F11-video-ledger-only.json",{"path":92,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":93,"sourceHtmlUrl":94},"evals\u002Ffixtures\u002Fgreen\u002F12-video-inspected-digest.json",1995,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fevals\u002Ffixtures\u002Fgreen\u002F12-video-inspected-digest.json",{"path":96,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":97,"sourceHtmlUrl":98},"evals\u002Ffixtures\u002Fred\u002F01-fail-page-never-loaded.json",1247,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fevals\u002Ffixtures\u002Fred\u002F01-fail-page-never-loaded.json",{"path":100,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":101,"sourceHtmlUrl":102},"evals\u002Ffixtures\u002Fred\u002F02-complete-no-report.json",944,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fevals\u002Ffixtures\u002Fred\u002F02-complete-no-report.json",{"path":104,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":105,"sourceHtmlUrl":106},"evals\u002Ffixtures\u002Fred\u002F03-fail-no-observation.json",1249,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fevals\u002Ffixtures\u002Fred\u002F03-fail-no-observation.json",{"path":108,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":109,"sourceHtmlUrl":110},"evals\u002Ffixtures\u002Fred\u002F04-prose-bypass-observed-error.json",1035,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fevals\u002Ffixtures\u002Fred\u002F04-prose-bypass-observed-error.json",{"path":112,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":113,"sourceHtmlUrl":114},"evals\u002Ffixtures\u002Fred\u002F05-passive-read-supplies-error.json",1202,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fevals\u002Ffixtures\u002Fred\u002F05-passive-read-supplies-error.json",{"path":116,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":117,"sourceHtmlUrl":118},"evals\u002Ffixtures\u002Fred\u002F06-pass-unreachable.json",1415,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fevals\u002Ffixtures\u002Fred\u002F06-pass-unreachable.json",{"path":120,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":121,"sourceHtmlUrl":122},"evals\u002Ffixtures\u002Fred\u002F07-report-stub-one-item.json",1274,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fevals\u002Ffixtures\u002Fred\u002F07-report-stub-one-item.json",{"path":124,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":125,"sourceHtmlUrl":126},"evals\u002Ffixtures\u002Fred\u002F08-fail-prose-unreached.json",1004,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fevals\u002Ffixtures\u002Fred\u002F08-fail-prose-unreached.json",{"path":128,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":129,"sourceHtmlUrl":130},"evals\u002Ffixtures\u002Fred\u002F09-namespaced-write-fail-no-observation.json",1018,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fevals\u002Ffixtures\u002Fred\u002F09-namespaced-write-fail-no-observation.json",{"path":132,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":133,"sourceHtmlUrl":134},"evals\u002Ffixtures\u002Fred\u002F10-fail-navigate-only.json",1183,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fevals\u002Ffixtures\u002Fred\u002F10-fail-navigate-only.json",{"path":136,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":137,"sourceHtmlUrl":138},"evals\u002Ffixtures\u002Fred\u002F11-decision-claim-no-visual-reference.json",1585,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fevals\u002Ffixtures\u002Fred\u002F11-decision-claim-no-visual-reference.json",{"path":140,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":141,"sourceHtmlUrl":142},"evals\u002Ffixtures\u002Fred\u002F12-user-facing-ready-merge-only.json",1437,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fevals\u002Ffixtures\u002Fred\u002F12-user-facing-ready-merge-only.json",{"path":144,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":145,"sourceHtmlUrl":146},"evals\u002Ffixtures\u002Fred\u002F13-video-no-ledger.json",1501,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fevals\u002Ffixtures\u002Fred\u002F13-video-no-ledger.json",{"path":148,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":149,"sourceHtmlUrl":150},"evals\u002Ffixtures\u002Fred\u002F14-video-ledger-in-prose.json",1868,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fevals\u002Ffixtures\u002Fred\u002F14-video-ledger-in-prose.json",{"path":152,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":153,"sourceHtmlUrl":154},"evals\u002Ffixtures\u002Fred\u002F15-video-different-finding.json",1860,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fevals\u002Ffixtures\u002Fred\u002F15-video-different-finding.json",{"path":156,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":157,"sourceHtmlUrl":158},"evals\u002Ffixtures\u002Fred\u002F16-video-prior-turn.json",1950,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fevals\u002Ffixtures\u002Fred\u002F16-video-prior-turn.json",{"path":160,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":85,"sourceHtmlUrl":161},"evals\u002Ffixtures\u002Fred\u002F17-video-different-run.json","https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fevals\u002Ffixtures\u002Fred\u002F17-video-different-run.json",{"path":163,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":164,"sourceHtmlUrl":165},"evals\u002Ffixtures\u002Fred\u002F18-video-malformed-ledger.json",1861,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fevals\u002Ffixtures\u002Fred\u002F18-video-malformed-ledger.json",{"path":167,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":168,"sourceHtmlUrl":169},"evals\u002Ffixtures\u002Fred\u002F19-video-unobserved-second-finding.json",1919,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fevals\u002Ffixtures\u002Fred\u002F19-video-unobserved-second-finding.json",{"path":171,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":172,"sourceHtmlUrl":173},"evals\u002Ffixtures\u002Fred\u002F20-video-traversal-frame.json",1892,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fevals\u002Ffixtures\u002Fred\u002F20-video-traversal-frame.json",{"path":175,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":176,"sourceHtmlUrl":177},"evals\u002Ffixtures\u002Fred\u002F21-video-confirmation-only.json",1725,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fevals\u002Ffixtures\u002Fred\u002F21-video-confirmation-only.json",{"path":179,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":180,"sourceHtmlUrl":181},"evals\u002Ffixtures\u002Fred\u002F22-video-inspected-malformed-digest.json",1931,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fevals\u002Ffixtures\u002Fred\u002F22-video-inspected-malformed-digest.json",{"path":183,"resourceType":184,"resourceTypeDesc":185,"sizeBytes":186,"sourceHtmlUrl":187},"evals\u002Fqa-verdict-gate.test.mjs","SCRIPT","脚本",5257,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fevals\u002Fqa-verdict-gate.test.mjs",{"path":189,"resourceType":184,"resourceTypeDesc":185,"sizeBytes":190,"sourceHtmlUrl":191},"evals\u002Frun_suite.py",6860,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fevals\u002Frun_suite.py",{"path":193,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":194,"sourceHtmlUrl":195},"install-snippet.json",304,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Finstall-snippet.json",{"path":197,"resourceType":184,"resourceTypeDesc":185,"sizeBytes":198,"sourceHtmlUrl":199},"lib\u002Ftranscript.mjs",6880,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Flib\u002Ftranscript.mjs",{"path":201,"resourceType":184,"resourceTypeDesc":185,"sizeBytes":202,"sourceHtmlUrl":203},"scripts\u002Fqa-verdict-gate-cli.mjs",1653,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fscripts\u002Fqa-verdict-gate-cli.mjs",{"path":205,"resourceType":184,"resourceTypeDesc":185,"sizeBytes":206,"sourceHtmlUrl":207},"scripts\u002Fqa-verdict-gate-hook.mjs",1641,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fscripts\u002Fqa-verdict-gate-hook.mjs",{"path":209,"resourceType":184,"resourceTypeDesc":185,"sizeBytes":210,"sourceHtmlUrl":211},"src\u002Fqa-verdict-gate.mjs",25001,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fqa-verdict-gate\u002Fsrc\u002Fqa-verdict-gate.mjs",[213,220,227,234,241,248],{"skillNo":214,"name":215,"nameZh":6,"description":216,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":217,"categories":218,"keywords":219,"publishedAt":19},"01M7C53JEF6A6SS20Q0D2YJYM8","yash","Push dependency bugs upstream. Triggers: dependency bug, workaround, monkey-patch, fork\u002Ffix. NOT app bugs.","https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fyash\u002FSKILL.md",[],[],{"skillNo":221,"name":222,"nameZh":6,"description":223,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":224,"categories":225,"keywords":226,"publishedAt":19},"01M7C53EPMNSZKHK47VY7WQDP0","whats-new","Diff Claude\u002FCodex\u002FCursor\u002FWispr\u002Fagy changelogs vs local setup. Triggers: whats new, changelog, version bump.","https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fwhats-new\u002FSKILL.md",[],[],{"skillNo":228,"name":229,"nameZh":6,"description":230,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":231,"categories":232,"keywords":233,"publishedAt":19},"01M7C53C4TVQ9G7AB3X57DACPB","weave","Orchestrator convergence: mine JSONLs, cite findings, route actions. Triggers: weave, run weave.","https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fweave\u002FSKILL.md",[],[],{"skillNo":235,"name":236,"nameZh":6,"description":237,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":238,"categories":239,"keywords":240,"publishedAt":19},"01M7C5342EDDN7YH53KKW511V2","unslop","Cut AI tells while keeping every receipt, number, path, and hedge intact. Triggers: unslop, cut the slop, wall of text, too verbose, rambly, make this readable. NOT for delivery cadence (see i-have-adhd), channel choice, or artifact length governance.","https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Funslop\u002FSKILL.md",[],[],{"skillNo":242,"name":243,"nameZh":6,"description":244,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":245,"categories":246,"keywords":247,"publishedAt":19},"01M7C52M0CTDW5DDP4452FJV0G","tmp-block","Guard denying durable writes to \u002Ftmp, \u002Fprivate\u002Ftmp, \u002Fvar\u002Ffolders, $TMPDIR; enforces \u003Crepo>\u002F.worktrees\u002F\u003Cname>. Triggers: tmp guard, TMP-BLOCK deny, WORKTREE-CONVENTION deny, bypass ledger. NOT for storage policy.","https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Ftmp-block\u002FSKILL.md",[],[],{"skillNo":249,"name":250,"nameZh":6,"description":251,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":252,"categories":253,"keywords":254,"publishedAt":19},"01M7C52HN4Q9E1Z74Z11K23E2H","t3-visuals","Show\u002Fvisualize\u002Fchart\u002Fdiagram\u002Fcompare\u002Fstatus board in T3 with compact interactive HTML. Not standalone dashboards.","https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Ft3-visuals\u002FSKILL.md",[],[]]