[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M6D33WSPRS0TTQFSEC2EKGKD":3,"skill-same:01M6D33WSPRS0TTQFSEC2EKGKD":76},{"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":26,"publishedAt":43,"linkOnly":44,"markdownContent":45,"markdownBody":46,"frontMatter":47,"refName":52,"commitSha":53,"skillDirectory":54,"forksCount":55,"repositoryHtmlUrl":56,"repositoryHomepageUrl":57,"ownerHtmlUrl":58,"ownerBlogUrl":59,"ownerDescription":60,"updatedAt":61,"install":62,"resources":75},"01M6D33WSPRS0TTQFSEC2EKGKD","benchmark-test-skill",null,"Run verify and benchmark tests for one agentic-skills skill, producing pass-rate, latency, cost, and consistency metrics","GeorgeQLe\u002Fagentic-skills","GeorgeQLe","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F21134126?v=4","01M6D2JP418S0VGJGF8M93EC3H","George Le","01M6D2JPBAKYD82TA5FS6HKQ3D",1,"MIT","https:\u002F\u002Fgithub.com\u002FGeorgeQLe\u002Fagentic-skills\u002Fblob\u002Fmaster\u002Fpacks\u002Fagentic-skills-bench\u002Fclaude\u002Fbenchmark-test-skill\u002Farchive\u002Fv0.4\u002FSKILL.md",[18,23],{"categoryNo":19,"categoryName":20,"taxonomyType":21,"taxonomyTypeDesc":22},"01JBSEEDCAT000000000000008","测试与质量","SCENARIO","使用场景",{"categoryNo":24,"categoryName":25,"taxonomyType":21,"taxonomyTypeDesc":22},"01JBSEEDCAT000000000000023","Skill与智能体开发",[27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42],"test","Test Skill","benchmark","benchmark test skill","agentic","run","metrics","execution","verify","skill name","page","v0","cost","alignment","latency","tests","2026-11-10 10:12:30",false,"---\nname: benchmark-test-skill\ndescription: Run verify and benchmark tests for one agentic-skills skill, producing pass-rate, latency, cost, and consistency metrics\ntype: execution\nversion: v0.4\nrequired_conventions: [alignment-page]\nargument-hint: \"\u003Cskill name>\"\n---\n\n# Benchmark Test Skill\n\nInvoke as `\u002Fbenchmark-test-skill \u003Cskill>`.\n\nUse this skill when the user wants to benchmark-test a skill from the public `agentic-skills` catalog using the separate `agentic-skills-benchmarks` repository. The trailing argument is the skill under test, not a mode for that skill. For example, `\u002Fbenchmark-test-skill design-system` tests the `design-system` skill with the benchmark harness; it does not run `design-system` against the current app or website.\n\nRun the benchmark repo verification gate followed by the benchmark extension for a single skill. By default, benchmark both Claude and Codex runners and report them separately.\n\nProduce deterministic benchmark evidence only. It should hand off to `\u002Fbenchmark-agent-review \u003Cskill>` as a separate step when the user needs subjective ergonomic judgment or remediation planning for the generated skill outputs.\n\n## Input\n\n- Required: one skill name, such as `design-system`.\n- If no skill name is provided, ask the user which skill to benchmark-test.\n- When invoked as `\u002Fbenchmark-test-skill \u003Cskill>`, resolve `benchmark-test-skill` as the active command first, including this project-local pack path, before treating the trailing argument as the skill under test.\n- Do not reinterpret the trailing argument as the active workflow unless the benchmark-test-skill command cannot be found after checking project-local packs.\n\n## Execution\n\nRun commands from `\u002FUsers\u002Fgeorgele\u002Fprojects\u002Ftools\u002Fagentic-skills-benchmarks` (or the local checkout of the public `agentic-skills-benchmarks` repo). If the checkout is missing, stop and ask for its path instead of running benchmark commands from `agentic-skills`.\n\n### Step 0 - Command Resolution Guard\n\nConfirm the active workflow is this `packs\u002Fagentic-skills-bench` skill. The requested target skill is data for the benchmark harness, not a command to execute directly.\n\n- For command-like invocations, preserve the leading command and route by the pack command first.\n- If a target such as `design-system`, `run`, or `ship` is provided, never run that skill directly as the benchmark action.\n- If command resolution is ambiguous, stop and report the ambiguity instead of running the target skill.\n\n### Step 1 - Eligibility Preflight\n\nBefore running verify, check whether the requested skill is a repository skill known to the benchmark harness:\n\n```bash\nSKILLS_REPO_URL=\u002FUsers\u002Fgeorgele\u002Fprojects\u002Ftools\u002Fagentic-skills SKILLS_REPO_REF=WORKTREE pnpm catalog:check\npnpm bench -- --list-skills\n```\n\n- If `\u003CSKILL>` is not listed, stop immediately and report `unknown skill: \u003CSKILL>`.\n- List the known repository skills from the command output.\n- Do not run `pnpm verify` or `pnpm bench` for unknown skills.\n- Read and report the listed coverage status for `\u003CSKILL>`: `custom`, `generic`, or `blocked`.\n- Skills with custom layer4 setups use skill-specific fixtures and hard assertions. Some setups also include deterministic output-quality rubrics.\n- Skills without custom layer4 setups use the harness generic smoke benchmark. Treat that as invocation\u002Fcompliance evidence, not deep domain-quality evidence.\n- If the row is `blocked`, stop before verify and bench. Report the blocked reason and next command from the list output.\n- If the row is `generic`, continue only as generic smoke evidence and route missing custom coverage to `\u002Ftargeted-skill-builder \u003CSKILL> benchmark coverage`.\n\n### Step 2 - Verify\n\n```bash\nSKILLS_REPO_URL=\u002FUsers\u002Fgeorgele\u002Fprojects\u002Ftools\u002Fagentic-skills SKILLS_REPO_REF=WORKTREE pnpm verify -- --skill \u003CSKILL>\n```\n\n- Expect layer1 to pass and layer2 to pass when target-specific tests exist.\n- Treat layer1 as the static harness-contract gate, including basic benchmark setup alignment checks such as expected next-route handoffs, runner command conventions, output file expectations, and quality-rubric facts against the current skill contract.\n- If layer1 fails because a benchmark setup is misaligned with the skill contract, classify it as a harness\u002Fbenchmark coverage defect and route to `\u002Ftargeted-skill-builder \u003CSKILL> benchmark failure`; do not spend agent budget on `pnpm bench`.\n- If layer2 reports no tests matched `\u003CSKILL>`, treat that layer as skipped and continue to the benchmark step. Record the skip clearly because generic benchmark coverage is weaker than target-specific layer2 verification.\n- Record pass\u002Ffail and wall time per layer.\n- If verify fails, stop and report the failure. Do not run the benchmark step.\n\n### Step 3 - Bench\n\nRun only if verify passes:\n\n```bash\npnpm bench -- --skill \u003CSKILL> --agent both --runs 3 --chunk-size 3 --pause 0\n```\n\n- Use 3 iterations by default.\n- Use `--agent both` by default. Only use `--agent claude` or `--agent codex` when the user explicitly asks to isolate one runner.\n- The expected budget is about $1 per run for the current design-system variant; report actual cost from the benchmark output when available.\n- The bench system persists raw data inside `agentic-skills-benchmarks` at `tests\u002Fbenchmarks\u002Fruns\u002F\u003Cskill>-\u003Cagent>-\u003CsessionId>\u002F` and generates `report.json`.\n- Treat rate limits, quota exhaustion, and similar runner-capacity errors as infrastructure-blocked runs, not skill failures. Report them separately from evaluated pass rate.\n- If recent same-skill benchmark reports or triage reports show repeated same-family benchmark false negatives, do not recommend another blind rerun as the next step. Report the recurrence and route to `\u002Ftargeted-skill-builder \u003CSKILL> benchmark repeated false-negative generalization` so the harness\u002Fsetup gets a family-level semantic evaluator, fixture set, or infrastructure classifier instead of another one-off tolerance patch.\n\n### Step 3.5 - Regression Check\n\nRun only if the bench step produced an evaluated report (skip if every run was infrastructure-blocked). This closes the benchmark loop by comparing the fresh grade to the prior grade in `benchmark\u002Fgrade-history.json` in the benchmark repo:\n\n```bash\nnode scripts\u002Fbenchmark-regression-check.mjs \u003CSKILL>\n```\n\n- The script reads the newest `report.json` per agent, appends the new grade to `benchmark\u002Fgrade-history.json` in `agentic-skills-benchmarks`, and prints an `improvement` \u002F `stable` \u002F `regression` verdict per agent.\n- A `regression` verdict (passRate, Wilson lower bound, or output-quality drop >= 10pp, or a status-badge demotion) is distinct from an absolute benchmark failure: the skill may still pass its thresholds yet score materially worse than its last graded run. The script exits non-zero on regression.\n- On a `regression` verdict, carry the printed prior-vs-new delta block into the next step and route to `\u002Fsession-triage \u003CSKILL> benchmark regression` so a human decides whether it is a real behavioral regression or harness\u002Frubric drift. Do not auto-fix.\n- On `improvement` or `stable`, continue to the report; note the verdict in the report.\n- See `docs\u002Fbenchmark-improvement-loop.md` for the full human-approved cycle this step participates in.\n\n### Step 4 - Report\n\nWrite results to `benchmark\u002Ftest-\u003CSKILL>-\u003CYYYY-MM-DD>.md` in `agentic-skills-benchmarks`. Use the current date.\n\nPopulate the report from `report.json` and verify the output includes:\n\n- verify table with layer status and wall time\n- agent name, evaluated pass rate, blocked-run count, and Wilson 95% confidence interval\n- failed assertions, if any\n- output-quality score summary when the setup defines a quality evaluator, including threshold failures, critical failures, and lowest-scoring criteria when present\n- infrastructure-blocked runs, if any\n- regression verdict from Step 3.5 (`improvement` \u002F `stable` \u002F `regression`) with the prior-vs-new delta when a prior grade existed\n- latency p50, p95, and p99\n- cost per run and total cost\n- mean pairwise similarity and outlier count\n- raw session path\n- recommended next route, using a literal label such as `Recommended next command:` or `Recommended next skill:`\n\nAfter writing the report, verify the file exists and contains the benchmark target, agent rows, pass-rate or blocked-run data, latency, cost, and raw session path. If any required report field is missing, treat the workflow as incomplete and fix the report before marking the benchmark done.\n\n## Output\n\nPrint a concise benchmark summary:\n\n- verify pass\u002Ffail\n- hard assertion pass rate\n- output-quality score when present, labeled as an additional rubric score rather than a statistical confidence measure\n- infrastructure-blocked count, if any\n- p50 latency\n- total cost\n- report path\n- next review handoff when subjective output-quality judgment or remediation planning is still needed\n\nThe Markdown report and the final assistant response must both include a literal next-route label accepted by the harness, such as `Recommended next skill: \u002Fbenchmark-agent-review \u003Cskill>` or `Recommended next command: \u002Fship`.\n\n## Alignment Page\n\nFollow the shared alignment-page convention via the packaged convention resolver; output path is `alignment\u002Fbenchmark-test-skill-{topic}.html`. By default, report results inline and write only this skill's normal durable artifacts; create an alignment page only when explicitly requested or when a concrete clarification\u002Freview need cannot be handled cleanly inline.\n\n## Constraints\n\n- Do not audit or benchmark the app, website, docs, or product surface unless the user explicitly asks for a separate website\u002Fproduct benchmark workflow.\n- Do not run `pnpm bench` when `pnpm verify` fails.\n- Do not fabricate benchmark metrics. Use the command output and `report.json`.\n- Do not present quality score as a replacement for hard assertion pass rate, or present a small benchmark run as statistically definitive.\n- Do not omit the final next-step route. Completion output must include either `Recommended next skill: \u003Ccommand>` or the two-line pair `**Next work:** \u003Cspecific task or \"none\">` and `**Recommended next command:** \u003Cone command or route>`.\n- Do not create or modify GitHub Actions workflows.\n\n## Next-Step Routing\n\nIf the skill is unknown to the repository, recommend checking the skill name or creating the skill first with `\u002Fcreate-agentic-skill`.\n\nIf the skill has blocked benchmark coverage, recommend the row's `next_command`.\n\nIf the skill only has generic smoke benchmark coverage or otherwise lacks custom domain-quality assertions, recommend `\u002Ftargeted-skill-builder \u003Cskill> benchmark coverage`.\n\nIf the skill fails verification, hard benchmark assertions, or configured quality thresholds, recommend `\u002Fsession-triage \u003Cskill> benchmark failure`.\n\nIf `scripts\u002Fbenchmark-regression-check.mjs` reports a `regression` verdict (the skill still grades but materially worse than its last graded run), recommend `\u002Fsession-triage \u003Cskill> benchmark regression` and carry the prior-vs-new delta block into that handoff.\n\nIf benchmark runs are blocked only by rate limits or quota exhaustion, recommend re-running `\u002Fbenchmark-test-skill \u003Cskill>` after the reset instead of treating the skill as failed.\n\nIf the failure pattern has already been triaged as a repeated same-family benchmark false negative, recommend `\u002Ftargeted-skill-builder \u003Cskill> benchmark repeated false-negative generalization` instead of another `\u002Fbenchmark-test-skill \u003Cskill>` rerun.\n\nIf evaluated benchmark runs completed and subjective output-quality review or remediation planning has not yet been performed, recommend `\u002Fbenchmark-agent-review \u003Cskill>` as the next separate step.\n\nIf the skill passes, the report is written, and no subjective review is needed or the separate `\u002Fbenchmark-agent-review \u003Cskill>` step has already been completed, recommend `\u002Fship`.\n\n## Default Shipping Contract\n\nFollow the shared shipping contract convention in CLAUDE.md.\n","# Benchmark Test Skill\n\nInvoke as `\u002Fbenchmark-test-skill \u003Cskill>`.\n\nUse this skill when the user wants to benchmark-test a skill from the public `agentic-skills` catalog using the separate `agentic-skills-benchmarks` repository. The trailing argument is the skill under test, not a mode for that skill. For example, `\u002Fbenchmark-test-skill design-system` tests the `design-system` skill with the benchmark harness; it does not run `design-system` against the current app or website.\n\nRun the benchmark repo verification gate followed by the benchmark extension for a single skill. By default, benchmark both Claude and Codex runners and report them separately.\n\nProduce deterministic benchmark evidence only. It should hand off to `\u002Fbenchmark-agent-review \u003Cskill>` as a separate step when the user needs subjective ergonomic judgment or remediation planning for the generated skill outputs.\n\n## Input\n\n- Required: one skill name, such as `design-system`.\n- If no skill name is provided, ask the user which skill to benchmark-test.\n- When invoked as `\u002Fbenchmark-test-skill \u003Cskill>`, resolve `benchmark-test-skill` as the active command first, including this project-local pack path, before treating the trailing argument as the skill under test.\n- Do not reinterpret the trailing argument as the active workflow unless the benchmark-test-skill command cannot be found after checking project-local packs.\n\n## Execution\n\nRun commands from `\u002FUsers\u002Fgeorgele\u002Fprojects\u002Ftools\u002Fagentic-skills-benchmarks` (or the local checkout of the public `agentic-skills-benchmarks` repo). If the checkout is missing, stop and ask for its path instead of running benchmark commands from `agentic-skills`.\n\n### Step 0 - Command Resolution Guard\n\nConfirm the active workflow is this `packs\u002Fagentic-skills-bench` skill. The requested target skill is data for the benchmark harness, not a command to execute directly.\n\n- For command-like invocations, preserve the leading command and route by the pack command first.\n- If a target such as `design-system`, `run`, or `ship` is provided, never run that skill directly as the benchmark action.\n- If command resolution is ambiguous, stop and report the ambiguity instead of running the target skill.\n\n### Step 1 - Eligibility Preflight\n\nBefore running verify, check whether the requested skill is a repository skill known to the benchmark harness:\n\n```bash\nSKILLS_REPO_URL=\u002FUsers\u002Fgeorgele\u002Fprojects\u002Ftools\u002Fagentic-skills SKILLS_REPO_REF=WORKTREE pnpm catalog:check\npnpm bench -- --list-skills\n```\n\n- If `\u003CSKILL>` is not listed, stop immediately and report `unknown skill: \u003CSKILL>`.\n- List the known repository skills from the command output.\n- Do not run `pnpm verify` or `pnpm bench` for unknown skills.\n- Read and report the listed coverage status for `\u003CSKILL>`: `custom`, `generic`, or `blocked`.\n- Skills with custom layer4 setups use skill-specific fixtures and hard assertions. Some setups also include deterministic output-quality rubrics.\n- Skills without custom layer4 setups use the harness generic smoke benchmark. Treat that as invocation\u002Fcompliance evidence, not deep domain-quality evidence.\n- If the row is `blocked`, stop before verify and bench. Report the blocked reason and next command from the list output.\n- If the row is `generic`, continue only as generic smoke evidence and route missing custom coverage to `\u002Ftargeted-skill-builder \u003CSKILL> benchmark coverage`.\n\n### Step 2 - Verify\n\n```bash\nSKILLS_REPO_URL=\u002FUsers\u002Fgeorgele\u002Fprojects\u002Ftools\u002Fagentic-skills SKILLS_REPO_REF=WORKTREE pnpm verify -- --skill \u003CSKILL>\n```\n\n- Expect layer1 to pass and layer2 to pass when target-specific tests exist.\n- Treat layer1 as the static harness-contract gate, including basic benchmark setup alignment checks such as expected next-route handoffs, runner command conventions, output file expectations, and quality-rubric facts against the current skill contract.\n- If layer1 fails because a benchmark setup is misaligned with the skill contract, classify it as a harness\u002Fbenchmark coverage defect and route to `\u002Ftargeted-skill-builder \u003CSKILL> benchmark failure`; do not spend agent budget on `pnpm bench`.\n- If layer2 reports no tests matched `\u003CSKILL>`, treat that layer as skipped and continue to the benchmark step. Record the skip clearly because generic benchmark coverage is weaker than target-specific layer2 verification.\n- Record pass\u002Ffail and wall time per layer.\n- If verify fails, stop and report the failure. Do not run the benchmark step.\n\n### Step 3 - Bench\n\nRun only if verify passes:\n\n```bash\npnpm bench -- --skill \u003CSKILL> --agent both --runs 3 --chunk-size 3 --pause 0\n```\n\n- Use 3 iterations by default.\n- Use `--agent both` by default. Only use `--agent claude` or `--agent codex` when the user explicitly asks to isolate one runner.\n- The expected budget is about $1 per run for the current design-system variant; report actual cost from the benchmark output when available.\n- The bench system persists raw data inside `agentic-skills-benchmarks` at `tests\u002Fbenchmarks\u002Fruns\u002F\u003Cskill>-\u003Cagent>-\u003CsessionId>\u002F` and generates `report.json`.\n- Treat rate limits, quota exhaustion, and similar runner-capacity errors as infrastructure-blocked runs, not skill failures. Report them separately from evaluated pass rate.\n- If recent same-skill benchmark reports or triage reports show repeated same-family benchmark false negatives, do not recommend another blind rerun as the next step. Report the recurrence and route to `\u002Ftargeted-skill-builder \u003CSKILL> benchmark repeated false-negative generalization` so the harness\u002Fsetup gets a family-level semantic evaluator, fixture set, or infrastructure classifier instead of another one-off tolerance patch.\n\n### Step 3.5 - Regression Check\n\nRun only if the bench step produced an evaluated report (skip if every run was infrastructure-blocked). This closes the benchmark loop by comparing the fresh grade to the prior grade in `benchmark\u002Fgrade-history.json` in the benchmark repo:\n\n```bash\nnode scripts\u002Fbenchmark-regression-check.mjs \u003CSKILL>\n```\n\n- The script reads the newest `report.json` per agent, appends the new grade to `benchmark\u002Fgrade-history.json` in `agentic-skills-benchmarks`, and prints an `improvement` \u002F `stable` \u002F `regression` verdict per agent.\n- A `regression` verdict (passRate, Wilson lower bound, or output-quality drop >= 10pp, or a status-badge demotion) is distinct from an absolute benchmark failure: the skill may still pass its thresholds yet score materially worse than its last graded run. The script exits non-zero on regression.\n- On a `regression` verdict, carry the printed prior-vs-new delta block into the next step and route to `\u002Fsession-triage \u003CSKILL> benchmark regression` so a human decides whether it is a real behavioral regression or harness\u002Frubric drift. Do not auto-fix.\n- On `improvement` or `stable`, continue to the report; note the verdict in the report.\n- See `docs\u002Fbenchmark-improvement-loop.md` for the full human-approved cycle this step participates in.\n\n### Step 4 - Report\n\nWrite results to `benchmark\u002Ftest-\u003CSKILL>-\u003CYYYY-MM-DD>.md` in `agentic-skills-benchmarks`. Use the current date.\n\nPopulate the report from `report.json` and verify the output includes:\n\n- verify table with layer status and wall time\n- agent name, evaluated pass rate, blocked-run count, and Wilson 95% confidence interval\n- failed assertions, if any\n- output-quality score summary when the setup defines a quality evaluator, including threshold failures, critical failures, and lowest-scoring criteria when present\n- infrastructure-blocked runs, if any\n- regression verdict from Step 3.5 (`improvement` \u002F `stable` \u002F `regression`) with the prior-vs-new delta when a prior grade existed\n- latency p50, p95, and p99\n- cost per run and total cost\n- mean pairwise similarity and outlier count\n- raw session path\n- recommended next route, using a literal label such as `Recommended next command:` or `Recommended next skill:`\n\nAfter writing the report, verify the file exists and contains the benchmark target, agent rows, pass-rate or blocked-run data, latency, cost, and raw session path. If any required report field is missing, treat the workflow as incomplete and fix the report before marking the benchmark done.\n\n## Output\n\nPrint a concise benchmark summary:\n\n- verify pass\u002Ffail\n- hard assertion pass rate\n- output-quality score when present, labeled as an additional rubric score rather than a statistical confidence measure\n- infrastructure-blocked count, if any\n- p50 latency\n- total cost\n- report path\n- next review handoff when subjective output-quality judgment or remediation planning is still needed\n\nThe Markdown report and the final assistant response must both include a literal next-route label accepted by the harness, such as `Recommended next skill: \u002Fbenchmark-agent-review \u003Cskill>` or `Recommended next command: \u002Fship`.\n\n## Alignment Page\n\nFollow the shared alignment-page convention via the packaged convention resolver; output path is `alignment\u002Fbenchmark-test-skill-{topic}.html`. By default, report results inline and write only this skill's normal durable artifacts; create an alignment page only when explicitly requested or when a concrete clarification\u002Freview need cannot be handled cleanly inline.\n\n## Constraints\n\n- Do not audit or benchmark the app, website, docs, or product surface unless the user explicitly asks for a separate website\u002Fproduct benchmark workflow.\n- Do not run `pnpm bench` when `pnpm verify` fails.\n- Do not fabricate benchmark metrics. Use the command output and `report.json`.\n- Do not present quality score as a replacement for hard assertion pass rate, or present a small benchmark run as statistically definitive.\n- Do not omit the final next-step route. Completion output must include either `Recommended next skill: \u003Ccommand>` or the two-line pair `**Next work:** \u003Cspecific task or \"none\">` and `**Recommended next command:** \u003Cone command or route>`.\n- Do not create or modify GitHub Actions workflows.\n\n## Next-Step Routing\n\nIf the skill is unknown to the repository, recommend checking the skill name or creating the skill first with `\u002Fcreate-agentic-skill`.\n\nIf the skill has blocked benchmark coverage, recommend the row's `next_command`.\n\nIf the skill only has generic smoke benchmark coverage or otherwise lacks custom domain-quality assertions, recommend `\u002Ftargeted-skill-builder \u003Cskill> benchmark coverage`.\n\nIf the skill fails verification, hard benchmark assertions, or configured quality thresholds, recommend `\u002Fsession-triage \u003Cskill> benchmark failure`.\n\nIf `scripts\u002Fbenchmark-regression-check.mjs` reports a `regression` verdict (the skill still grades but materially worse than its last graded run), recommend `\u002Fsession-triage \u003Cskill> benchmark regression` and carry the prior-vs-new delta block into that handoff.\n\nIf benchmark runs are blocked only by rate limits or quota exhaustion, recommend re-running `\u002Fbenchmark-test-skill \u003Cskill>` after the reset instead of treating the skill as failed.\n\nIf the failure pattern has already been triaged as a repeated same-family benchmark false negative, recommend `\u002Ftargeted-skill-builder \u003Cskill> benchmark repeated false-negative generalization` instead of another `\u002Fbenchmark-test-skill \u003Cskill>` rerun.\n\nIf evaluated benchmark runs completed and subjective output-quality review or remediation planning has not yet been performed, recommend `\u002Fbenchmark-agent-review \u003Cskill>` as the next separate step.\n\nIf the skill passes, the report is written, and no subjective review is needed or the separate `\u002Fbenchmark-agent-review \u003Cskill>` step has already been completed, recommend `\u002Fship`.\n\n## Default Shipping Contract\n\nFollow the shared shipping contract convention in CLAUDE.md.\n",{"name":5,"type":34,"version":48,"description":7,"argument-hint":49,"required_conventions":50},"v0.4","\u003Cskill name>",[51],"alignment-page","master","96e2912af886dd1606b1b97df707715aef062b02","packs\u002Fagentic-skills-bench\u002Fclaude\u002Fbenchmark-test-skill\u002Farchive\u002Fv0.4",0,"https:\u002F\u002Fgithub.com\u002FGeorgeQLe\u002Fagentic-skills","https:\u002F\u002Fagentic-skills-blush.vercel.app","https:\u002F\u002Fgithub.com\u002FGeorgeQLe","georgeqle.com","George Le is a futurist and serial tech entrepreneur who wants to work toward a world where humanity is free from robotic tasks.","2026-11-10 04:07:24",{"source":8,"skillName":5,"commands":63,"prompt":73,"zipAvailable":74},[64,67,70],{"tool":65,"command":66},"npx","npx skills add GeorgeQLe\u002Fagentic-skills --skill benchmark-test-skill",{"tool":68,"command":69},"bunx","bunx skills add GeorgeQLe\u002Fagentic-skills --skill benchmark-test-skill",{"tool":71,"command":72},"pnpm","pnpm dlx skills add GeorgeQLe\u002Fagentic-skills --skill benchmark-test-skill","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FGeorgeQLe\u002Fagentic-skills\u002Fblob\u002Fmaster\u002Fpacks\u002Fagentic-skills-bench\u002Fclaude\u002Fbenchmark-test-skill\u002Farchive\u002Fv0.4\u002FSKILL.md",true,[],[77,89,99,111,122,133],{"skillNo":78,"name":79,"nameZh":6,"description":80,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":81,"categories":82,"keywords":83,"publishedAt":43},"01M6G0NJES15RJKZE6YZ0TSYGF","source-checkout-allowed","Valid source-checkout-only fixture.","https:\u002F\u002Fgithub.com\u002FGeorgeQLe\u002Fagentic-skills\u002Fblob\u002Fmaster\u002Ftests\u002Ffixtures\u002Fskill-install-routing\u002Fvalid\u002Fsource-checkout-allowed\u002FSKILL.md",[],[84,85,86,87,38,88],"source checkout allowed","checkout","fixture","only","valid",{"skillNo":90,"name":91,"nameZh":6,"description":92,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":93,"categories":94,"keywords":95,"publishedAt":43},"01M6G0NFZ3R1MNB1E3DVTM0WB7","deck-install-valid","Valid deck install fixture.","https:\u002F\u002Fgithub.com\u002FGeorgeQLe\u002Fagentic-skills\u002Fblob\u002Fmaster\u002Ftests\u002Ffixtures\u002Fskill-install-routing\u002Fvalid\u002Fdeck-install\u002FSKILL.md",[],[96,97,98,88,86,38],"deck install valid","deck","install",{"skillNo":100,"name":101,"nameZh":6,"description":102,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":103,"categories":104,"keywords":105,"publishedAt":43},"01M6G0ND3QZQC4MDZ9NVB97PT2","codex-pack-guard-valid","Valid Codex Pack Availability Guard fixture.","https:\u002F\u002Fgithub.com\u002FGeorgeQLe\u002Fagentic-skills\u002Fblob\u002Fmaster\u002Ftests\u002Ffixtures\u002Fskill-install-routing\u002Fvalid\u002Fcodex-pack-guard\u002FSKILL.md",[],[106,107,108,88,109,110,86,38],"codex pack guard valid","pack","guard","codex","availability",{"skillNo":112,"name":113,"nameZh":6,"description":114,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":115,"categories":116,"keywords":118,"publishedAt":43},"01M6G0NA9YB3NKGZKD3S96JMNB","codex-missing-skill-valid","Valid Codex missing skill fallback fixture.","https:\u002F\u002Fgithub.com\u002FGeorgeQLe\u002Fagentic-skills\u002Fblob\u002Fmaster\u002Ftests\u002Ffixtures\u002Fskill-install-routing\u002Fvalid\u002Fcodex-missing-skill-fallback\u002FSKILL.md",[117],{"categoryNo":24,"categoryName":25,"taxonomyType":21,"taxonomyTypeDesc":22},[119,120,88,109,38,121,86],"codex missing skill valid","missing","fallback",{"skillNo":123,"name":124,"nameZh":6,"description":125,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":126,"categories":127,"keywords":129,"publishedAt":43},"01M6G0N7PC0EKFD40ZWQAXSAQ9","codex-base-skill-fallback-valid","Valid Codex base skill fallback fixture.","https:\u002F\u002Fgithub.com\u002FGeorgeQLe\u002Fagentic-skills\u002Fblob\u002Fmaster\u002Ftests\u002Ffixtures\u002Fskill-install-routing\u002Fvalid\u002Fcodex-base-skill-fallback\u002FSKILL.md",[128],{"categoryNo":24,"categoryName":25,"taxonomyType":21,"taxonomyTypeDesc":22},[88,109,121,130,131,132,38,86],"base skill","codex base skill fallback valid","base",{"skillNo":134,"name":135,"nameZh":6,"description":136,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":137,"categories":138,"keywords":139,"publishedAt":43},"01M6G0N4VRR2YKZQ6N3Y6S2BQA","claude-pack-guard-valid","Valid Claude Pack Availability Guard fixture.","https:\u002F\u002Fgithub.com\u002FGeorgeQLe\u002Fagentic-skills\u002Fblob\u002Fmaster\u002Ftests\u002Ffixtures\u002Fskill-install-routing\u002Fvalid\u002Fclaude-pack-guard\u002FSKILL.md",[],[88,108,107,140,110,38,86],"claude pack guard valid"]