[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7FF8S763J0VM4GHTWEPCM2J":3,"skill-same:01M7FF8S763J0VM4GHTWEPCM2J":52},{"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":24,"commitSha":25,"skillDirectory":26,"forksCount":27,"repositoryHtmlUrl":28,"repositoryHomepageUrl":6,"ownerHtmlUrl":29,"ownerBlogUrl":6,"ownerDescription":30,"updatedAt":31,"install":32,"resources":45},"01M7FF8S763J0VM4GHTWEPCM2J","loopx-benchmark",null,"Use when a LoopX-managed goal runs, tracks, scores, or analyzes a benchmark experiment through benchmark-toolkit, including experiment-board rows, solver arms, integrity qualification, matched comparisons, or case insights. Do not use for casual benchmark discussion, ordinary software microbenchmarks, or eval mentions without LoopX experiment state.","lza6\u002FArmor-piercing-system-prompts","lza6","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F128880206?v=4","01M7EB3K7WPX5T28EJ2NVHE6EG","Chinese-tingfeng","01M7EB3K97E1KQ16QWKJMFNGX8",11,"MIT","https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F07-后端与架构\u002Floopx-benchmark\u002FSKILL.md",[],[],"2026-11-16 10:05:34",false,"---\nname: loopx-benchmark\ndescription: Use when a LoopX-managed goal runs, tracks, scores, or analyzes a benchmark experiment through benchmark-toolkit, including experiment-board rows, solver arms, integrity qualification, matched comparisons, or case insights. Do not use for casual benchmark discussion, ordinary software microbenchmarks, or eval mentions without LoopX experiment state.\n---\n\n# LoopX Benchmark Workflow\n\nUse this skill for a LoopX-managed benchmark experiment. The builtin\n`benchmark-toolkit` capability owns provider-neutral experiment state and\nintegrity boundaries. This packaged skill is its task-triggered Agent playbook.\n\nThe capability is catalog-ready without a per-Goal enable switch. Installing\nthis skill does not grant runner, shell, network, credential, private-evidence,\nor Goal mutation authority. Respect the selected todo's required capabilities,\nany external provider binding, host permissions, and user gates.\n\n## Capability surface\n\n- `loopx capability show benchmark-toolkit --format json` — catalog entry with\n  usage hints, role boundaries, and the post-run case-insight template.\n- `loopx benchmark --help` — subcommands (experiment-board-show,\n  experiment-board-upsert, source-revision-fence, integrity-qualification,\n  classify-artifacts).\n\n## Share a study through the public-safe contract\n\nWhen another benchmark developer needs portable study data, use the capability's\ntyped study flow rather than sharing a runner-specific ledger or raw evidence:\n\n1. Validate `benchmark_study_manifest_v0` with `benchmark study-validate`.\n2. Wrap one allowlisted manifest, experiment-board row, redacted insight, or runtime\n   observation with `benchmark upload-envelope`.\n3. Run `benchmark upload-local` without `--execute` first, then explicitly execute\n   against a caller-selected local JSONL store.\n4. Verify the record\u002Fdigest\u002Frevision binding with `benchmark upload-readback`.\n5. Derive the campaign\u002Farm\u002Fcase\u002Frun packet with `benchmark study-dashboard`; pass a\n   compact four-arm contract only when the study preregistered that design.\n\nFor `case_insight_projection`, first upload the same run's active terminal\nexperiment-board row with `insight.status=complete`. The case, run, and outcome\nmust match; the run row remains the only arm, score, countability, integrity, and\ntreatment-fidelity authority. Reduce private post-run evidence to bounded prose\nand public-safe handles or digests before building the envelope.\n\nThe local provider is a no-network simulation. It does not grant remote upload,\npublication, credentials, retention, or benchmark submission authority. Adapters\nkeep their native metric names and reduce private post-run evidence before envelope\nconstruction.\n\n## Share exploratory behavior findings\n\nWhen the owner authorizes selected behavioral observations but not a complete\nstudy release, use `behavior_finding` records and `benchmark behavior-report`.\nSee `docs\u002Freference\u002Fbenchmark-behavior-findings.md` for the contract. These records\nrequire selection rules, sample denominators, observations, interpretations,\nlimitations, counterevidence, and evidence digests; they require neither a run-row\nupload nor a full study manifest and have no score authority.\n\nFreeze the authorized disclosure projection before rendering. Review the same\nscope in visible text, foldouts, embedded data, downloads, and PR attachments.\nPermission to share duration does not grant permission to share outcome totals\nor deltas. Schema validity and a producer redaction attestation are not publication\napproval or verification of unshared evidence. Keep selected-case observations\nexplicitly exploratory and retain the relevant limitations and counterexamples.\n\n## Select the operating lane\n\n- **Inspect or explain:** use `capability show` and `benchmark --help`; remain\n  read-only. Do not create an experiment-board row merely because the user asks\n  what the toolkit does.\n- **Plan, select, or launch a run:** follow the experiment sequence below. The\n  first action is to read the board; a launch still requires an authorized\n  runner and admitted source.\n- **Monitor an active campaign:** read the board and runtime-owned projections;\n  update only on material run transitions. Do not manufacture progress from a\n  timer tick.\n- **Analyze a terminal run:** wait until solving is terminal and scoring is\n  complete before reading hidden evaluator evidence or writing a case insight.\n\nFor a generic library microbenchmark or an eval with no LoopX Goal\u002Fboard, use\nthe task's normal tools instead of imposing this workflow.\n\n## Experiment sequence\n\n1. **Read the experiment board before launching or selecting a case.**\n   ```bash\n   loopx benchmark experiment-board-show --goal-id \u003CGOAL_ID> --format json\n   ```\n   Inspect baseline, treatment, explore, countability, effort, and insight rows\n   before choosing the next arm.\n\n2. **Qualify the source revision before each new run admission.**\n   ```bash\n   loopx benchmark source-revision-fence \\\n     --source-checkout \u003Cclean-source> \\\n     --expected-revision \u003CPIN> \\\n     --observed-reference-revision \u003COBSERVED_HEAD> \\\n     --require-admitted --format json\n   ```\n   The fence fails closed unless the clean pinned source matches the observed\n   reference head.\n\n3. **Preview, then preregister or mark the run row when it starts.**\n   ```bash\n   loopx benchmark experiment-board-upsert --goal-id \u003CGOAL_ID> \\\n     --row-json \u003Crunning-row.json> --format json\n   loopx benchmark experiment-board-upsert --goal-id \u003CGOAL_ID> \\\n     --row-json \u003Crunning-row.json> --execute --format json\n   ```\n   The running row uses `status=running`, empty `metrics`, and\n   `countability={integrity_qualified:false, official_result_present:false,\n   score_countable:false}`. Keep the same stable `run_id` for every transition.\n\n4. **Preview and upsert terminal score, countability, effort, and insight.**\n   First run integrity qualification. An automated restricted-access match is a\n   countable suspicion, not a cheating verdict. After solver and scoring are\n   terminal, inspect the real solver trajectory, tool results, and final\n   workspace. Pass a compact\n   `benchmark_restricted_access_adjudication_v0` only after that review; confirm\n   cheating only when restricted material was actually disclosed and causally\n   entered a solving or validation decision.\n\n   ```bash\n   loopx benchmark experiment-board-upsert --goal-id \u003CGOAL_ID> \\\n     --row-json \u003Cterminal-row.json> --execute --format json\n   ```\n   The terminal row sets `status=completed`, fills `metrics` (primary metric plus\n   guardrails), and updates `countability`. Only mark `score_countable=true` when\n   `integrity_qualified=true` and `official_result_present=true`. Fill `effort`\n   and set `insight.status` to `complete` after the post-run analysis.\n\n   For non-baseline arms, also reduce the reviewed mechanism facts separately:\n   ```bash\n   loopx benchmark treatment-continuation-receipt \\\n     --observation-json \u003Ccompact-post-run-observation.json> --format json\n   ```\n   This receipt distinguishes qualified startup from post-start semantic control\n   persistence. It is analysis-only and must not change score countability,\n   integrity qualification, treatment fidelity, or matched-pair eligibility.\n\n5. **Read matched comparisons before selecting the next arm.**\n   ```bash\n   loopx benchmark experiment-board-show --goal-id \u003CGOAL_ID> --format json\n   ```\n   Only claim paired results from `matched_pair_countable` comparisons. Keep\n   diagnostic-only explore rows in a separate evidence lane.\n\n## Run-row contract\n\n- `benchmark_id`, `study_id`, `case_id`, `run_id`, `arm_id`, `arm_role`,\n  `attempt`, `status`, `observed_at`, `model_id`, `protocol_id`,\n  `comparison_protocol_id`, `claim_scope`, `primary_metric`,\n  `guardrail_metrics`, `metrics`, `countability`, `treatment_fidelity`,\n  `effort`, `insight` are the canonical row fields (`schema_version` =\n  `benchmark_experiment_board_row_v0`).\n- Baseline rows must use `treatment_fidelity=not_applicable` and cannot name a\n  `comparison_anchor_run_id`. Non-baseline rows must name a\n  `comparison_anchor_run_id`.\n- Metrics are `{\"name\": {\"value\": \u003Cnumber>, \"unit\": \u003Cstr>,\n  \"higher_is_better\": \u003Cbool>}}`; at most 16 entries. The `primary_metric` must\n  not also be a guardrail metric.\n- `score_countable` requires `status=completed`, `integrity_qualified=true`,\n  and `official_result_present=true`. `score=0` is a valid completed result.\n\n## Source, integrity, and artifact boundaries\n\n- `source-revision-fence` is read-only and caller-observed: it performs no\n  fetch, install, or launch. It blocks new admissions only.\n- `integrity-qualification` reduces private trajectory and runner isolation\n  evidence to a compact public-safe receipt (hashes, counts, reason codes).\n- Scanner hits for restricted source access or host-boundary escape probes set\n  `restricted_access_review=suspected` while keeping the run score-eligible. Use\n  `--restricted-access-adjudication-json` for the post-run agent decision; only\n  confirmed disclosure plus causal use disqualifies the score.\n- `classify-artifacts` classifies benchmark artifact paths without reading them;\n  use it before reading or publishing any candidate artifact.\n- The solver lane must not read hidden tests, verifier sources, gold answers, or\n  official feedback during the solving phase. The post-run analyst may read full\n  private evidence only after the solver is terminal and scoring is complete.\n- `capability bind` selects an external provider implementation for a Goal; it\n  is not the activation mechanism for this builtin capability. Todo\n  `required_capability` fields remain runtime prerequisites, not product\n  capability switches.\n\n## Campaign monitoring and post-run insight\n\n- When a campaign starts and the caller authorizes ongoing monitoring, add one\n  `continuous_monitor` todo. Refresh aggregate score\u002Fcoverage and write\n  `benchmark_case_insight_v0` on material scored-case transitions, with bounded\n  periodic reviews while the campaign remains active.\n- Treat that monitor as an observation lane, not executable delivery. When a\n  material poll discovers bounded repository, runner-repair, or experiment work,\n  use `quota monitor-poll --material-change --next-agent-todo` with explicit\n  `--next-action-kind`, repository, and required capabilities so it creates an\n  independent runnable `advancement_task`. An unchanged poll creates no successor\n  and spends no delivery quota.\n- If the main campaign advancement Todo is waiting for a monitor transition, keep\n  it `open` and pair `resume_when=monitor_changed:\u003Cmonitor-todo-id>` with an\n  already-created independent runnable successor. Do not mark the wait `blocked`,\n  and do not treat the monitor itself as delivery work.\n- Report only public-safe conclusions (countable baselines, countable\n  treatments, matched pairs, aggregate primary metric by arm, improved\u002Fflat\u002F\n  regressed pair counts). Never copy raw private evidence into a user update.\n- After a solver stops and scoring completes, read the task, real trajectory,\n  final workspace, hidden tests, verifier, and failure\u002Fscore details; write one\n  `benchmark_case_insight_v0` explaining the decisive evidence, why the outcome\n  happened, and what LoopX should test next.\n- For treatment arms, record whether qualified startup was followed by semantic\n  Todo transitions, technical replans, or control closeout. Use `startup_only`\n  only when a complete authorized post-run review observed no such transition;\n  otherwise absence is `unknown`. Keep terminal settlement separate.\n- Do not send a repetitive user update when nothing material changed.\n","# LoopX Benchmark Workflow\n\nUse this skill for a LoopX-managed benchmark experiment. The builtin\n`benchmark-toolkit` capability owns provider-neutral experiment state and\nintegrity boundaries. This packaged skill is its task-triggered Agent playbook.\n\nThe capability is catalog-ready without a per-Goal enable switch. Installing\nthis skill does not grant runner, shell, network, credential, private-evidence,\nor Goal mutation authority. Respect the selected todo's required capabilities,\nany external provider binding, host permissions, and user gates.\n\n## Capability surface\n\n- `loopx capability show benchmark-toolkit --format json` — catalog entry with\n  usage hints, role boundaries, and the post-run case-insight template.\n- `loopx benchmark --help` — subcommands (experiment-board-show,\n  experiment-board-upsert, source-revision-fence, integrity-qualification,\n  classify-artifacts).\n\n## Share a study through the public-safe contract\n\nWhen another benchmark developer needs portable study data, use the capability's\ntyped study flow rather than sharing a runner-specific ledger or raw evidence:\n\n1. Validate `benchmark_study_manifest_v0` with `benchmark study-validate`.\n2. Wrap one allowlisted manifest, experiment-board row, redacted insight, or runtime\n   observation with `benchmark upload-envelope`.\n3. Run `benchmark upload-local` without `--execute` first, then explicitly execute\n   against a caller-selected local JSONL store.\n4. Verify the record\u002Fdigest\u002Frevision binding with `benchmark upload-readback`.\n5. Derive the campaign\u002Farm\u002Fcase\u002Frun packet with `benchmark study-dashboard`; pass a\n   compact four-arm contract only when the study preregistered that design.\n\nFor `case_insight_projection`, first upload the same run's active terminal\nexperiment-board row with `insight.status=complete`. The case, run, and outcome\nmust match; the run row remains the only arm, score, countability, integrity, and\ntreatment-fidelity authority. Reduce private post-run evidence to bounded prose\nand public-safe handles or digests before building the envelope.\n\nThe local provider is a no-network simulation. It does not grant remote upload,\npublication, credentials, retention, or benchmark submission authority. Adapters\nkeep their native metric names and reduce private post-run evidence before envelope\nconstruction.\n\n## Share exploratory behavior findings\n\nWhen the owner authorizes selected behavioral observations but not a complete\nstudy release, use `behavior_finding` records and `benchmark behavior-report`.\nSee `docs\u002Freference\u002Fbenchmark-behavior-findings.md` for the contract. These records\nrequire selection rules, sample denominators, observations, interpretations,\nlimitations, counterevidence, and evidence digests; they require neither a run-row\nupload nor a full study manifest and have no score authority.\n\nFreeze the authorized disclosure projection before rendering. Review the same\nscope in visible text, foldouts, embedded data, downloads, and PR attachments.\nPermission to share duration does not grant permission to share outcome totals\nor deltas. Schema validity and a producer redaction attestation are not publication\napproval or verification of unshared evidence. Keep selected-case observations\nexplicitly exploratory and retain the relevant limitations and counterexamples.\n\n## Select the operating lane\n\n- **Inspect or explain:** use `capability show` and `benchmark --help`; remain\n  read-only. Do not create an experiment-board row merely because the user asks\n  what the toolkit does.\n- **Plan, select, or launch a run:** follow the experiment sequence below. The\n  first action is to read the board; a launch still requires an authorized\n  runner and admitted source.\n- **Monitor an active campaign:** read the board and runtime-owned projections;\n  update only on material run transitions. Do not manufacture progress from a\n  timer tick.\n- **Analyze a terminal run:** wait until solving is terminal and scoring is\n  complete before reading hidden evaluator evidence or writing a case insight.\n\nFor a generic library microbenchmark or an eval with no LoopX Goal\u002Fboard, use\nthe task's normal tools instead of imposing this workflow.\n\n## Experiment sequence\n\n1. **Read the experiment board before launching or selecting a case.**\n   ```bash\n   loopx benchmark experiment-board-show --goal-id \u003CGOAL_ID> --format json\n   ```\n   Inspect baseline, treatment, explore, countability, effort, and insight rows\n   before choosing the next arm.\n\n2. **Qualify the source revision before each new run admission.**\n   ```bash\n   loopx benchmark source-revision-fence \\\n     --source-checkout \u003Cclean-source> \\\n     --expected-revision \u003CPIN> \\\n     --observed-reference-revision \u003COBSERVED_HEAD> \\\n     --require-admitted --format json\n   ```\n   The fence fails closed unless the clean pinned source matches the observed\n   reference head.\n\n3. **Preview, then preregister or mark the run row when it starts.**\n   ```bash\n   loopx benchmark experiment-board-upsert --goal-id \u003CGOAL_ID> \\\n     --row-json \u003Crunning-row.json> --format json\n   loopx benchmark experiment-board-upsert --goal-id \u003CGOAL_ID> \\\n     --row-json \u003Crunning-row.json> --execute --format json\n   ```\n   The running row uses `status=running`, empty `metrics`, and\n   `countability={integrity_qualified:false, official_result_present:false,\n   score_countable:false}`. Keep the same stable `run_id` for every transition.\n\n4. **Preview and upsert terminal score, countability, effort, and insight.**\n   First run integrity qualification. An automated restricted-access match is a\n   countable suspicion, not a cheating verdict. After solver and scoring are\n   terminal, inspect the real solver trajectory, tool results, and final\n   workspace. Pass a compact\n   `benchmark_restricted_access_adjudication_v0` only after that review; confirm\n   cheating only when restricted material was actually disclosed and causally\n   entered a solving or validation decision.\n\n   ```bash\n   loopx benchmark experiment-board-upsert --goal-id \u003CGOAL_ID> \\\n     --row-json \u003Cterminal-row.json> --execute --format json\n   ```\n   The terminal row sets `status=completed`, fills `metrics` (primary metric plus\n   guardrails), and updates `countability`. Only mark `score_countable=true` when\n   `integrity_qualified=true` and `official_result_present=true`. Fill `effort`\n   and set `insight.status` to `complete` after the post-run analysis.\n\n   For non-baseline arms, also reduce the reviewed mechanism facts separately:\n   ```bash\n   loopx benchmark treatment-continuation-receipt \\\n     --observation-json \u003Ccompact-post-run-observation.json> --format json\n   ```\n   This receipt distinguishes qualified startup from post-start semantic control\n   persistence. It is analysis-only and must not change score countability,\n   integrity qualification, treatment fidelity, or matched-pair eligibility.\n\n5. **Read matched comparisons before selecting the next arm.**\n   ```bash\n   loopx benchmark experiment-board-show --goal-id \u003CGOAL_ID> --format json\n   ```\n   Only claim paired results from `matched_pair_countable` comparisons. Keep\n   diagnostic-only explore rows in a separate evidence lane.\n\n## Run-row contract\n\n- `benchmark_id`, `study_id`, `case_id`, `run_id`, `arm_id`, `arm_role`,\n  `attempt`, `status`, `observed_at`, `model_id`, `protocol_id`,\n  `comparison_protocol_id`, `claim_scope`, `primary_metric`,\n  `guardrail_metrics`, `metrics`, `countability`, `treatment_fidelity`,\n  `effort`, `insight` are the canonical row fields (`schema_version` =\n  `benchmark_experiment_board_row_v0`).\n- Baseline rows must use `treatment_fidelity=not_applicable` and cannot name a\n  `comparison_anchor_run_id`. Non-baseline rows must name a\n  `comparison_anchor_run_id`.\n- Metrics are `{\"name\": {\"value\": \u003Cnumber>, \"unit\": \u003Cstr>,\n  \"higher_is_better\": \u003Cbool>}}`; at most 16 entries. The `primary_metric` must\n  not also be a guardrail metric.\n- `score_countable` requires `status=completed`, `integrity_qualified=true`,\n  and `official_result_present=true`. `score=0` is a valid completed result.\n\n## Source, integrity, and artifact boundaries\n\n- `source-revision-fence` is read-only and caller-observed: it performs no\n  fetch, install, or launch. It blocks new admissions only.\n- `integrity-qualification` reduces private trajectory and runner isolation\n  evidence to a compact public-safe receipt (hashes, counts, reason codes).\n- Scanner hits for restricted source access or host-boundary escape probes set\n  `restricted_access_review=suspected` while keeping the run score-eligible. Use\n  `--restricted-access-adjudication-json` for the post-run agent decision; only\n  confirmed disclosure plus causal use disqualifies the score.\n- `classify-artifacts` classifies benchmark artifact paths without reading them;\n  use it before reading or publishing any candidate artifact.\n- The solver lane must not read hidden tests, verifier sources, gold answers, or\n  official feedback during the solving phase. The post-run analyst may read full\n  private evidence only after the solver is terminal and scoring is complete.\n- `capability bind` selects an external provider implementation for a Goal; it\n  is not the activation mechanism for this builtin capability. Todo\n  `required_capability` fields remain runtime prerequisites, not product\n  capability switches.\n\n## Campaign monitoring and post-run insight\n\n- When a campaign starts and the caller authorizes ongoing monitoring, add one\n  `continuous_monitor` todo. Refresh aggregate score\u002Fcoverage and write\n  `benchmark_case_insight_v0` on material scored-case transitions, with bounded\n  periodic reviews while the campaign remains active.\n- Treat that monitor as an observation lane, not executable delivery. When a\n  material poll discovers bounded repository, runner-repair, or experiment work,\n  use `quota monitor-poll --material-change --next-agent-todo` with explicit\n  `--next-action-kind`, repository, and required capabilities so it creates an\n  independent runnable `advancement_task`. An unchanged poll creates no successor\n  and spends no delivery quota.\n- If the main campaign advancement Todo is waiting for a monitor transition, keep\n  it `open` and pair `resume_when=monitor_changed:\u003Cmonitor-todo-id>` with an\n  already-created independent runnable successor. Do not mark the wait `blocked`,\n  and do not treat the monitor itself as delivery work.\n- Report only public-safe conclusions (countable baselines, countable\n  treatments, matched pairs, aggregate primary metric by arm, improved\u002Fflat\u002F\n  regressed pair counts). Never copy raw private evidence into a user update.\n- After a solver stops and scoring completes, read the task, real trajectory,\n  final workspace, hidden tests, verifier, and failure\u002Fscore details; write one\n  `benchmark_case_insight_v0` explaining the decisive evidence, why the outcome\n  happened, and what LoopX should test next.\n- For treatment arms, record whether qualified startup was followed by semantic\n  Todo transitions, technical replans, or control closeout. Use `startup_only`\n  only when a complete authorized post-run review observed no such transition;\n  otherwise absence is `unknown`. Keep terminal settlement separate.\n- Do not send a repetitive user update when nothing material changed.\n",{"name":5,"description":7},"main","d69096e239f2a89aabc2cd078c47664a9d3b1e11","07-后端与架构\u002Floopx-benchmark",3,"https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts","https:\u002F\u002Fgithub.com\u002Flza6","个人vx：Tf00798\r\n个人✈：@TFGY999","2026-10-08 14:29:22",{"source":8,"skillName":5,"commands":33,"prompt":43,"zipAvailable":44},[34,37,40],{"tool":35,"command":36},"npx","npx skills add lza6\u002FArmor-piercing-system-prompts --skill loopx-benchmark",{"tool":38,"command":39},"bunx","bunx skills add lza6\u002FArmor-piercing-system-prompts --skill loopx-benchmark",{"tool":41,"command":42},"pnpm","pnpm dlx skills add lza6\u002FArmor-piercing-system-prompts --skill loopx-benchmark","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F07-后端与架构\u002Floopx-benchmark\u002FSKILL.md",true,[46],{"path":47,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":50,"sourceHtmlUrl":51},"agents\u002Fopenai.yaml","REFERENCE","参考文档",364,"https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F07-后端与架构\u002Floopx-benchmark\u002Fagents\u002Fopenai.yaml",[53,60,65,72,79,86],{"skillNo":54,"name":55,"nameZh":6,"description":56,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":57,"categories":58,"keywords":59,"publishedAt":19},"01M7FFG485KY83TD6HSC4YPCMC","obsidian-clipper-template-creator","Guide for creating templates for the Obsidian Web Clipper. Use when you want to create a new clipping template, understand available variables, or format clipped content.","https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F07-后端与架构\u002Fobsidian-clipper-template-creator\u002FSKILL.md",[],[],{"skillNo":61,"name":55,"nameZh":6,"description":56,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":62,"categories":63,"keywords":64,"publishedAt":19},"01M7FFG0T6AZ0STEJPSB3CX84K","https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F07-后端与架构\u002Fobsidian-clipper-template-creator-antigravity-awesome-skills-main\u002FSKILL.md",[],[],{"skillNo":66,"name":67,"nameZh":6,"description":68,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":69,"categories":70,"keywords":71,"publishedAt":19},"01M7FFFY0QEM14SKWPDYGMXM2P","python-api","Programmatic Object-Detection-Metrics API guidance for BoundingBox, BoundingBoxes, Evaluator, enums, coordinate conversion, AP\u002FmAP semantics, and source-style import diagnostics.","https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F07-后端与架构\u002Fobject-detection-metrics\u002Fsub-skills\u002Fpython-api\u002FSKILL.md",[],[],{"skillNo":73,"name":74,"nameZh":6,"description":75,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":76,"categories":77,"keywords":78,"publishedAt":19},"01M7FFFVZQZ2GJ0C0TGP7JHD20","file-evaluation","Folder-based PASCAL VOC AP\u002FmAP evaluation from ground-truth and detection text files.","https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F07-后端与架构\u002Fobject-detection-metrics\u002Fsub-skills\u002Ffile-evaluation\u002FSKILL.md",[],[],{"skillNo":80,"name":81,"nameZh":6,"description":82,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":83,"categories":84,"keywords":85,"publishedAt":19},"01M7FFFSFNYSE77SJ6PE1CK2DA","object-detection-metrics","Guides legacy Object-Detection-Metrics PASCAL VOC AP\u002FmAP evaluation from detection text folders and direct Python API classes.","https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F07-后端与架构\u002Fobject-detection-metrics\u002FSKILL.md",[],[],{"skillNo":87,"name":88,"nameZh":6,"description":89,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":90,"categories":91,"keywords":92,"publishedAt":19},"01M7FFFP0GCK1RVXBCPQ8APBMT","nx-workspace-patterns","Configure and optimize Nx monorepo workspaces. Use when setting up Nx, configuring project boundaries, optimizing build caching, or implementing affected commands.","https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F07-后端与架构\u002Fnx-workspace-patterns\u002FSKILL.md",[],[]]