[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M79NHREV79R3146AFMHVMHSN":3,"skill-same:01M79NHREV79R3146AFMHVMHSN":86},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":15,"categories":16,"keywords":17,"publishedAt":18,"linkOnly":19,"markdownContent":20,"markdownBody":21,"frontMatter":22,"refName":24,"commitSha":25,"skillDirectory":26,"forksCount":13,"repositoryHtmlUrl":27,"repositoryHomepageUrl":6,"ownerHtmlUrl":28,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":29,"install":30,"resources":43},"01M79NHREV79R3146AFMHVMHSN","skill-tester",null,"Test and benchmark Codex skills empirically using evaluation-driven development. Use when validating a skill's effectiveness, running evals, comparing skill vs. baseline performance, running benchmarks with timing\u002Ftoken metrics, or iterating on skill improvements based on empirical data. Supports both a fast pass\u002Ffail check and a full baseline-comparison benchmark with timing\u002Ftoken metrics.","AndreHahm\u002Fandres-cc-marketplace","AndreHahm","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F80748019?v=4","01M79N2VZJT9MAJ7K695C9VXQ4","01M79N2W06PFX8BSCQ592M8K8K",0,"Apache-2.0","https:\u002F\u002Fgithub.com\u002FAndreHahm\u002Fandres-cc-marketplace\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fskill-tester\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: skill-tester\ndescription: >-\n  Test and benchmark Codex skills empirically using evaluation-driven development.\n  Use when validating a skill's effectiveness, running evals, comparing skill vs. baseline\n  performance, running benchmarks with timing\u002Ftoken metrics, or iterating on skill improvements\n  based on empirical data. Supports both a fast pass\u002Ffail check and a full baseline-comparison\n  benchmark with timing\u002Ftoken metrics.\nallowed-tools: Read Write Edit Glob Grep Agent Bash(python:*) Skill\n---\n\n# Skill Tester\n\n**Purpose:** Empirically validate Codex skills through evaluation-driven testing. Proves skills actually help Codex (with data) rather than guessing.\n\n## Mindset\n\nSkills must be **measured, not assumed**. This pipeline provides systematic evidence: Does the skill improve Codex's performance? By how much? What should we improve next?\n\n## When to Use\n\n- Validating a newly created skill against baseline Codex performance without it\n- Benchmarking skill refinements across multiple iterations\n- Proving skill effectiveness with timing, token usage, and pass rates side-by-side\n- Making data-driven decisions on what to improve next in a skill\n\n## When NOT to Use\n\n- **Creating skills** — use `skill-development` instead.\n- **Reviewing skill quality** — use `skill-reviewer` for a single-pass structural quality check without benchmarking.\n- **Validating plugin structure** — use the `plugin-validator` agent for manifest and directory checks.\n- **Testing during initial skill authoring (Phase 3 of skill creation, before the skill is finalized)** — use `skill-development`'s own Phase 3 workflow; switch to this skill once the skill exists and you need a dedicated with\u002Fbaseline benchmark or a multi-iteration comparison.\n- **Automated structural fix-review loops until skill-reviewer passes** — use `skill-improver-loop` instead; this skill iterates on empirical eval\u002Fbenchmark data, not skill-reviewer's structural findings.\n\n## Core Principles\n\n**Empirical Evidence** — Intuition ≠ proof. Collect actual performance data: pass rates, tokens, timing.\n\n**Parallel Testing** — For each eval, test WITH skill AND baseline SIMULTANEOUSLY (2 agents per eval, in parallel). Eliminates confounding variables from different test runs.\n\n**Workspace Isolation** — Each iteration lives in `workspace\u002Fiteration-N\u002F`. Keeps history clean. Easy to compare iteration 1 → 2 → 3 performance.\n\n**Schema Consistency** — All JSON outputs follow strict schemas (evals.json, grading.json, benchmark.json, timing.json). Enables reliable aggregation and comparison.\n\n## Quick Start: The 7-Phase Pipeline\n\n```\nPhase 1: Setup          → Identify skill + confirm what it does\nPhase 2: Create Evals   → Interview user → write test cases + assertions\nPhase 3: Run Tests      → Launch 2 agents per eval (with_skill + baseline) in parallel\nPhase 4: Grade Results  → Evaluate outputs against assertions\nPhase 5: Aggregate      → Run aggregation script to compute benchmark.json\nPhase 6: Review Summary → Show comparison table + improvement suggestions\nPhase 7: Iterate        → Update skill + run next iteration (or stop if satisfied)\n```\n\n## Artifact Location (Standardized)\n\n**All evaluation artifacts live in a centralized `.\u002Fevals\u002F` directory at project root.**\n\n```\n.\u002Fevals\u002F                          ← Project root, NOT inside skill directory\n├── \u003Cskill-name-1>\u002F\n│   ├── evals.json\n│   └── workspace\u002F\n│       ├── iteration-1\u002F\n│       │   ├── eval-1\u002F, eval-2\u002F, eval-3\u002F  ← Per-eval directories\n│       │   └── benchmark.json              ← Aggregated results\n│       └── iteration-2\u002F\n│           └── ...\n└── \u003Cskill-name-2>\u002F\n    └── ...\n```\n\nSee `references\u002Fworkflow.md` — \"Phase 1: Setup\" for workspace location rationale.\n\n---\n\n## PHASE 1: Setup\n\n**Goal:** Identify the skill being tested. Confirm what it does. Prepare workspace.\n\n### Step 1.1: Ask User - Which Skill?\n\nUse AskUserQuestion to ask which skill to test:\n\n```\nquestion: \"Which skill do you want to test?\"\nheader: \"Skill Selection\"\noptions: [\n  {label: \"skill-development\", description: \"Testing skill-development from skills\u002F directory\"},\n  {label: \"skill-refiner-interactive\", description: \"Testing skill-refiner-interactive from skills\u002F directory\"},\n  {label: \"Other skill\", description: \"Testing a skill not listed above\"}\n]\n```\n\n⏸️ **Collect user input.** Wait for response before proceeding.\n\n### Step 1.2: Confirm Skill Path & Purpose\n\nRead the skill's SKILL.md to understand:\n- What does this skill do?\n- What are its core use cases?\n- What problem does it solve?\n\nShow user a summary:\n```\nSkill: \u003Cname>\nLocation: \u003Cpath>\nPurpose: \u003Cone-line summary from description>\n```\n\n### Step 1.2b: Choose Workflow Mode\n\n**Question:** Which testing mode do you want?\n\nUse AskUserQuestion with 2 options:\n\n```\nquestions: [\n  {\n    question: \"Which testing mode would you like?\",\n    header: \"Workflow Mode\",\n    options: [\n      {\n        label: \"Quick Workflow\",\n        description: \"Fast validation: only test WITH skill (no baseline), no timing metrics. Just pass\u002Ffail on assertions. Perfect for quick checks.\"\n      },\n      {\n        label: \"Full Pipeline\",\n        description: \"Complete analysis: test WITH skill + baseline parallel, measure tokens\u002Ftiming, aggregate & compare. For comprehensive benchmarking.\"\n      }\n    ],\n    multiSelect: false\n  }\n]\n```\n\n⏸️ **Collect user input.**\n\n**If Quick Workflow selected:**\n- Skip to Phase 2 with note: \"Quick mode: only with_skill, no baseline comparison\"\n- Jump to simplified Phase 3 (only 1 agent per eval)\n- Skip Phase 5 (aggregation) and Phase 6 comparison tables\n- Show simple pass\u002Ffail results\n\n**If Full Pipeline selected:**\n- Continue with standard 7-phase pipeline (proceed to Step 1.3)\n\n### Step 1.3: Create Workspace\n\nCreate directory structure in project root (standardized):\n```\n.\u002Fevals\u002F\u003Cskill-name>\u002F\n.\u002Fevals\u002F\u003Cskill-name>\u002Fworkspace\u002Fiteration-1\u002F\n```\n\nLog workspace path. All subsequent phases write to this directory (NOT inside the skill's directory).\n\n---\n\n## PHASE 2: Create Evals\n\n**Goal:** Build evaluation cases that measure skill effectiveness.\n\n### Step 2.1: Interview User for Test Scenarios\n\nAsk 3 questions plus an optional 4th (use AskUserQuestion with free-form \"Other\" option):\n\n```\nQuestion 1: \"What are 2-3 core scenarios this skill should handle?\"\nExample: \"skill-development should handle: creating a new skill from scratch,\nconverting a slash command, improving an existing skill\"\n\nQuestion 2: \"For each scenario, what makes a GOOD response?\"\nExample: \"Good responses: skill has clear name, description with trigger phrases,\ncorrect SKILL.md structure, efficient references\u002F\"\n\nQuestion 3: \"What should FAIL (baseline without skill)?\"\nExample: \"Baseline will miss best practices, create vague descriptions,\nskip necessary structure\"\n\nQuestion 4 (optional): \"Should any of these scenarios also be tested under pressure —\ne.g. time constraints, sunk cost, exhaustion, authority — to verify the skill holds up\nwhen an agent is incentivized to skip it?\"\nExample: \"Yes — the validation scenario should be re-run with a 'you have 5 minutes' framing\"\n```\n\n⏸️ **Collect responses.** Store in memory.\n\nIf Question 4 is answered yes, this pipeline's quantitative pass rates don't cover that axis — run `skill-development`'s Phase 3.5 compliance testing (`references\u002Fcompliance-testing.md`) before or alongside this pipeline; see \"Integration with skill-development, skill-refiner-interactive\" in `references\u002Fworkflow.md`.\n\n### Step 2.1b: Cross-Check Against the Target's Own Testing & Validation Section\n\nBefore writing `evals.json`, `Read` the target skill's own SKILL.md for a \"Testing & Validation\" section. If one exists and lists numbered scenarios, check whether the eval scenarios collected in Step 2.1 collectively exercise each item — or explicitly note the gap for any that aren't covered. A skill's own Testing & Validation checklist is a direct, authoritative claim about what \"tested\" should mean for it; writing an eval set that doesn't cross-check against it risks shipping eval coverage narrower than the skill's own documented claims, which reads as tested when it isn't.\n\n### Step 2.2: Generate evals.json\n\nWrite `.\u002Fevals\u002F\u003Cskill-name>\u002Fevals.json` with fields: `skill_name`, `skill_path`, `description`, and `evals` array (each entry: `id`, `name`, `prompt`, `expected_output`, `files`).\n\nSee `references\u002Feval-schema.md` for full schema and annotated examples.\n\n### Step 2.3: Generate eval_metadata.json Files\n\nCreate `workspace\u002Fiteration-1\u002Feval-N\u002Feval_metadata.json` for each eval. Required fields: `eval_id`, `skill_path`, `assertions` array (each entry: `text`, `type`, `target`). Assertion types: `presence`, `quality`, `structure`, `functionality`.\n\nCreate directories: `workspace\u002Fiteration-1\u002Feval-1\u002F`, `eval-2\u002F`, etc. (with_skill and baseline subdirs created in Phase 3).\n\nSee `references\u002Feval-schema.md` for full schema and annotated examples.\n\n### Step 2.4: Plugin-Rule Compliance Assertions (Optional)\n\nWhen evaluating a skill for plugin-rule compliance (naming, tool-scoping, language, formatting), invoke `plugin-rulebook` before writing `eval_metadata.json` to obtain the active rule set. Use each enabled rule as a separate assertion:\n\n```\nSkill: plugin-rulebook\n→ Returns active rules with descriptions, examples, and enforcement levels\n```\n\nAdd rule-compliance assertions to `eval_metadata.json` alongside functional assertions:\n\n```json\n{\"text\": \"allowed-tools uses space-separated format (R6)\", \"type\": \"structure\", \"target\": \"SKILL.md frontmatter\"}\n```\n\n---\n\n## QUICK WORKFLOW (Alternative Path)\n\n**If user chose \"Quick Workflow\" in Step 1.2b, follow this streamlined path instead of Phases 3-7.**\n\n### Quick Phase 1: Run WITH_SKILL Only (No Baseline)\n\n**Goal:** Test the skill itself. No baseline comparison, no timing metrics. Just: Does it work?\n\nFor EACH eval in evals.json, launch ONE agent (no parallel baseline):\n\nSee `references\u002Feval-schema.md` — \"Agent Prompt Templates\" section for the WITH_SKILL_ONLY prompt template (Quick Workflow variant, no timing metrics).\n\n⏸️ **Wait for agent completion before proceeding to next eval.**\n\n### Quick Phase 2: Grade Results (Assertions Only)\n\nFor each eval, grade against assertions in eval_metadata.json. Record pass\u002Ffail + evidence for each assertion.\nSave to: `.\u002Fevals\u002F\u003Cskill-name>\u002Fworkspace\u002Fiteration-1\u002Feval-N\u002Fwith_skill\u002Fgrading.json`\n\nSee `references\u002Feval-schema.md` for grading.json schema.\n\n### Quick Phase 3: Show Results\n\nDisplay simple pass\u002Ffail summary (no benchmark.json):\n\n```\nQUICK VALIDATION RESULTS: \u003Cskill-name>\n====================================\n\nEval 1: \u003CScenario>  ✓ PASS (5\u002F5 assertions)\nEval 2: \u003CScenario>  ✓ PASS (4\u002F5 assertions)\nEval 3: \u003CScenario>  ✗ FAIL (2\u002F5 assertions)\n\nSummary: 11\u002F15 assertions passed (73%)\nStatus: Ready to refine or deploy\n```\n\n### Quick Phase 4: Next Steps\n\nAsk user:\n\n```\nquestion: \"What would you like to do?\"\nheader: \"Next Steps\"\noptions: [\n  {label: \"Run full pipeline\", description: \"Move to complete benchmarking with baseline comparison\"},\n  {label: \"Refine skill\", description: \"Update skill based on failed assertions\"},\n  {label: \"Done\", description: \"Quick validation complete\"}\n]\n```\n\n---\n\n## PHASE 3: Run Tests\n\n**Goal:** Execute 2 agents per eval (WITH skill + BASELINE) in parallel. Capture outputs.\n\n### Step 3.1: Spawn Agents (Parallel Execution)\n\nFor EACH eval, launch 2 agents SIMULTANEOUSLY in one Agent tool call:\n\nSee `references\u002Feval-schema.md` — \"Agent Prompt Templates\" section for the WITH_SKILL and BASELINE prompt templates.\n\n⏸️ **Wait for both agents to complete before proceeding to Phase 4.**\n\n---\n\n## PHASE 4: Grade Results\n\n**Goal:** Evaluate each agent's outputs against assertions in eval_metadata.json.\n\n### Step 4.1: Review Outputs & Grade\n\nFor EACH eval:\n\n1. Read `with_skill\u002Foutputs\u002F` files\n2. Read `baseline\u002Foutputs\u002F` files\n3. For EACH assertion in `eval-N\u002Feval_metadata.json`:\n   - Does with_skill output satisfy the assertion? (true\u002Ffalse)\n   - Does baseline output satisfy the assertion? (true\u002Ffalse)\n   - Collect brief evidence quotes from outputs\n\n### Step 4.2: Write grading.json Files\n\nCreate `workspace\u002Fiteration-1\u002Feval-N\u002Fwith_skill\u002Fgrading.json` and `baseline\u002Fgrading.json`.\nRequired fields: `eval_id`, `configuration` (`\"with_skill\"` or `\"baseline\"`), `assertions_evaluated` array (each: `text`, `passed`, `evidence`), `summary` (`assertions_total`, `assertions_passed`, `pass_rate`).\n\nSee `references\u002Feval-schema.md` for full schema and pass\u002Ffail examples.\n\n---\n\n## PHASE 5: Aggregate\n\n**Goal:** Compute benchmark.json with summary stats (pass rates, tokens, timing).\n\n### Step 5.1: Run Aggregation Script\n\nExecute the aggregation script (see `scripts\u002Faggregate_benchmark.py` for the full implementation; `references\u002Feval-schema.md` for the invocation command and output schema):\n\n```bash\npython ${CLAUDE_SKILL_DIR}\u002Fscripts\u002Faggregate_benchmark.py \\\n  .\u002Fevals\u002F\u003Cskill-name>\u002Fworkspace\u002Fiteration-1\n```\n\nExample:\n```bash\npython ${CLAUDE_SKILL_DIR}\u002Fscripts\u002Faggregate_benchmark.py \\\n  .\u002Fevals\u002Fskill-development\u002Fworkspace\u002Fiteration-1\n```\n\nThis reads all grading.json and timing.json files from the standardized location, outputs `benchmark.json` with per-eval stats (pass rates, tokens, duration, delta) and summary statistics.\n\nSee `references\u002Feval-schema.md` for benchmark.json schema and full example.\n\n---\n\n## PHASE 6: Review Summary\n\n**Goal:** Show user a clear comparison table + next steps.\n\n### Step 6.1: Render Comparison Table\n\nDisplay results in human-readable format:\n\n```\nEVALUATION RESULTS: skill-tester iteration-1\n==============================================\n\nEval 1: [Scenario Name]\n  WITH SKILL:  3\u002F3 (100%) | 2500 tokens | 8s\n  BASELINE:    2\u002F3 (67%)  | 1800 tokens | 5s\n  DELTA:       +33% better | +700 tokens | +3s\n\nEval 2: [Scenario Name]\n  WITH SKILL:  2\u002F3 (67%)  | 2100 tokens | 7s\n  BASELINE:    2\u002F3 (67%)  | 1900 tokens | 4s\n  DELTA:       Even      | +200 tokens | +3s\n\nSUMMARY\n-------\nWith Skill Avg:    96.5% pass rate | 2300 avg tokens\nBaseline Avg:      67.0% pass rate | 1850 avg tokens\nIMPROVEMENT:       +29.5 percentage points\n```\n\n### Step 6.2: Offer Next Steps\n\nAsk user (AskUserQuestion):\n\n```\nquestion: \"What would you like to do?\"\nheader: \"Next Steps\"\noptions: [\n  {\n    label: \"Iterate (update skill, run next iteration)\",\n    description: \"Based on results, modify the skill and run workspace\u002Fiteration-2\"\n  },\n  {\n    label: \"Stop (satisfied with results)\",\n    description: \"Evaluation complete. Save benchmark data for reference\"\n  },\n  {\n    label: \"Refine evals (change test cases, rerun)\",\n    description: \"Update evals.json and run iteration again with same eval set\"\n  }\n]\n```\n\n⏸️ **Collect user input.**\n\n---\n\n## PHASE 7: Iterate\n\n**Goal:** If user chose \"Iterate\", update the skill and run next iteration.\n\n### Step 7.1: Get Improvement Feedback\n\nAsk user (AskUserQuestion, free-form):\n\n```\nquestion: \"What would you like to improve about the skill?\"\nheader: \"Improvement\"\n```\n\n⏸️ **Collect feedback.**\n\n### Step 7.2: Update Skill\n\nUse `skill-refiner-interactive` or direct edits to improve the skill based on feedback:\n- Reword descriptions for clarity\n- Add missing guidance to SKILL.md body\n- Reorganize references\n- Trim token usage\n\n### Step 7.3: Run Next Iteration\n\nCreate `workspace\u002Fiteration-2\u002F` directory. Repeat Phases 3–6 with updated skill.\n\n### Step 7.4: Compare Iterations\n\nShow delta between iteration-1 benchmark and iteration-2 benchmark:\n\n```\nITERATION COMPARISON\n====================\nIteration 1 pass rate: 67%  → Iteration 2: 95% (+28 points)\nIteration 1 tokens:   1900  → Iteration 2: 2100 (+200, acceptable)\n```\n\nLoop back to Phase 6 (Step 6.2) to ask: iterate again or stop?\n\n---\n\n**Suggested next step:** if a benchmark run shows a regression, a failed assertion, or a Quick Workflow FAIL, ask with `AskUserQuestion`: \"Run `enhancement-suggestor` against these results for a classified (complexity\u002Frisk\u002Fbenefit) WHAT\u002FWHY\u002FHOW action plan?\" — options \"Yes\" \u002F \"No\". If yes, invoke the `enhancement-suggestor` agent (via `Agent`) against the results. Never invoke it without asking first.\n\n## Testing & Validation\n\nAfter a test run, verify:\n\n1. **Trigger phrases** — confirm skill activates on: \"run evals on skill-X\", \"benchmark this skill\", \"validate skill performance\", \"compare skill vs. baseline\", \"test skill effectiveness\"\n2. **Non-triggers** — confirm skill does NOT activate on: \"review my skill structure\", \"create a new skill\", \"fix this PR\"\n3. **Mode selection** — Quick Workflow and Full Pipeline branches both produce correctly structured output directories\n4. **Schema integrity** — all JSON files (`evals.json`, `grading.json`, `benchmark.json`, `timing.json`, `eval_metadata.json`) validate against `references\u002Feval-schema.md` schemas\n5. **Baseline parity** — baseline agent receives no SKILL.md content; with_skill agent receives full SKILL.md content\n\n**Verified 2026-07-11:** items 1 and 2 checked directly against the current frontmatter description — each of the 5 trigger phrases maps to specific description language (\"running evals\", \"validating a skill's effectiveness\", \"comparing skill vs. baseline performance\", \"running benchmarks\"), and none of the 3 non-trigger phrases share that vocabulary (they map to `skill-reviewer`'s and `skill-development`'s domains instead). `aggregate_benchmark.py`'s new guard\u002Fsort behavior (items covering script robustness) was verified with synthetic fixtures — see the script's own commit history. A live end-to-end pipeline run (spawning real with_skill\u002Fbaseline agents) has not been performed — that remains the one unverified item.\n\n**Quality gates:**\n- [ ] WITH_SKILL and BASELINE agents launched simultaneously (not sequentially) in Full Pipeline\n- [ ] `benchmark.json` exists in `workspace\u002Fiteration-N\u002F` before presenting Phase 6 summary\n- [ ] All destructive operations (deleting eval dirs) confirmed with user before execution\n\n## Reference Guide\n\n| Resource | Purpose |\n|---|---|\n| `references\u002Feval-schema.md` | JSON schemas, agent prompt templates, and workspace structure |\n| `references\u002Fworkflow.md` | Decision points and detailed workflow guidance |\n| `scripts\u002Faggregate_benchmark.py` | Python script that aggregates grading\u002Ftiming data into benchmark.json |\n| `plugin-rulebook` skill | Active rule configuration for compliance-testing assertions |\n| `enhancement-suggestor` agent | Turns a regression\u002Ffail\u002Fdelta result into a classified, prioritized action plan |\n","# Skill Tester\n\n**Purpose:** Empirically validate Codex skills through evaluation-driven testing. Proves skills actually help Codex (with data) rather than guessing.\n\n## Mindset\n\nSkills must be **measured, not assumed**. This pipeline provides systematic evidence: Does the skill improve Codex's performance? By how much? What should we improve next?\n\n## When to Use\n\n- Validating a newly created skill against baseline Codex performance without it\n- Benchmarking skill refinements across multiple iterations\n- Proving skill effectiveness with timing, token usage, and pass rates side-by-side\n- Making data-driven decisions on what to improve next in a skill\n\n## When NOT to Use\n\n- **Creating skills** — use `skill-development` instead.\n- **Reviewing skill quality** — use `skill-reviewer` for a single-pass structural quality check without benchmarking.\n- **Validating plugin structure** — use the `plugin-validator` agent for manifest and directory checks.\n- **Testing during initial skill authoring (Phase 3 of skill creation, before the skill is finalized)** — use `skill-development`'s own Phase 3 workflow; switch to this skill once the skill exists and you need a dedicated with\u002Fbaseline benchmark or a multi-iteration comparison.\n- **Automated structural fix-review loops until skill-reviewer passes** — use `skill-improver-loop` instead; this skill iterates on empirical eval\u002Fbenchmark data, not skill-reviewer's structural findings.\n\n## Core Principles\n\n**Empirical Evidence** — Intuition ≠ proof. Collect actual performance data: pass rates, tokens, timing.\n\n**Parallel Testing** — For each eval, test WITH skill AND baseline SIMULTANEOUSLY (2 agents per eval, in parallel). Eliminates confounding variables from different test runs.\n\n**Workspace Isolation** — Each iteration lives in `workspace\u002Fiteration-N\u002F`. Keeps history clean. Easy to compare iteration 1 → 2 → 3 performance.\n\n**Schema Consistency** — All JSON outputs follow strict schemas (evals.json, grading.json, benchmark.json, timing.json). Enables reliable aggregation and comparison.\n\n## Quick Start: The 7-Phase Pipeline\n\n```\nPhase 1: Setup          → Identify skill + confirm what it does\nPhase 2: Create Evals   → Interview user → write test cases + assertions\nPhase 3: Run Tests      → Launch 2 agents per eval (with_skill + baseline) in parallel\nPhase 4: Grade Results  → Evaluate outputs against assertions\nPhase 5: Aggregate      → Run aggregation script to compute benchmark.json\nPhase 6: Review Summary → Show comparison table + improvement suggestions\nPhase 7: Iterate        → Update skill + run next iteration (or stop if satisfied)\n```\n\n## Artifact Location (Standardized)\n\n**All evaluation artifacts live in a centralized `.\u002Fevals\u002F` directory at project root.**\n\n```\n.\u002Fevals\u002F                          ← Project root, NOT inside skill directory\n├── \u003Cskill-name-1>\u002F\n│   ├── evals.json\n│   └── workspace\u002F\n│       ├── iteration-1\u002F\n│       │   ├── eval-1\u002F, eval-2\u002F, eval-3\u002F  ← Per-eval directories\n│       │   └── benchmark.json              ← Aggregated results\n│       └── iteration-2\u002F\n│           └── ...\n└── \u003Cskill-name-2>\u002F\n    └── ...\n```\n\nSee `references\u002Fworkflow.md` — \"Phase 1: Setup\" for workspace location rationale.\n\n---\n\n## PHASE 1: Setup\n\n**Goal:** Identify the skill being tested. Confirm what it does. Prepare workspace.\n\n### Step 1.1: Ask User - Which Skill?\n\nUse AskUserQuestion to ask which skill to test:\n\n```\nquestion: \"Which skill do you want to test?\"\nheader: \"Skill Selection\"\noptions: [\n  {label: \"skill-development\", description: \"Testing skill-development from skills\u002F directory\"},\n  {label: \"skill-refiner-interactive\", description: \"Testing skill-refiner-interactive from skills\u002F directory\"},\n  {label: \"Other skill\", description: \"Testing a skill not listed above\"}\n]\n```\n\n⏸️ **Collect user input.** Wait for response before proceeding.\n\n### Step 1.2: Confirm Skill Path & Purpose\n\nRead the skill's SKILL.md to understand:\n- What does this skill do?\n- What are its core use cases?\n- What problem does it solve?\n\nShow user a summary:\n```\nSkill: \u003Cname>\nLocation: \u003Cpath>\nPurpose: \u003Cone-line summary from description>\n```\n\n### Step 1.2b: Choose Workflow Mode\n\n**Question:** Which testing mode do you want?\n\nUse AskUserQuestion with 2 options:\n\n```\nquestions: [\n  {\n    question: \"Which testing mode would you like?\",\n    header: \"Workflow Mode\",\n    options: [\n      {\n        label: \"Quick Workflow\",\n        description: \"Fast validation: only test WITH skill (no baseline), no timing metrics. Just pass\u002Ffail on assertions. Perfect for quick checks.\"\n      },\n      {\n        label: \"Full Pipeline\",\n        description: \"Complete analysis: test WITH skill + baseline parallel, measure tokens\u002Ftiming, aggregate & compare. For comprehensive benchmarking.\"\n      }\n    ],\n    multiSelect: false\n  }\n]\n```\n\n⏸️ **Collect user input.**\n\n**If Quick Workflow selected:**\n- Skip to Phase 2 with note: \"Quick mode: only with_skill, no baseline comparison\"\n- Jump to simplified Phase 3 (only 1 agent per eval)\n- Skip Phase 5 (aggregation) and Phase 6 comparison tables\n- Show simple pass\u002Ffail results\n\n**If Full Pipeline selected:**\n- Continue with standard 7-phase pipeline (proceed to Step 1.3)\n\n### Step 1.3: Create Workspace\n\nCreate directory structure in project root (standardized):\n```\n.\u002Fevals\u002F\u003Cskill-name>\u002F\n.\u002Fevals\u002F\u003Cskill-name>\u002Fworkspace\u002Fiteration-1\u002F\n```\n\nLog workspace path. All subsequent phases write to this directory (NOT inside the skill's directory).\n\n---\n\n## PHASE 2: Create Evals\n\n**Goal:** Build evaluation cases that measure skill effectiveness.\n\n### Step 2.1: Interview User for Test Scenarios\n\nAsk 3 questions plus an optional 4th (use AskUserQuestion with free-form \"Other\" option):\n\n```\nQuestion 1: \"What are 2-3 core scenarios this skill should handle?\"\nExample: \"skill-development should handle: creating a new skill from scratch,\nconverting a slash command, improving an existing skill\"\n\nQuestion 2: \"For each scenario, what makes a GOOD response?\"\nExample: \"Good responses: skill has clear name, description with trigger phrases,\ncorrect SKILL.md structure, efficient references\u002F\"\n\nQuestion 3: \"What should FAIL (baseline without skill)?\"\nExample: \"Baseline will miss best practices, create vague descriptions,\nskip necessary structure\"\n\nQuestion 4 (optional): \"Should any of these scenarios also be tested under pressure —\ne.g. time constraints, sunk cost, exhaustion, authority — to verify the skill holds up\nwhen an agent is incentivized to skip it?\"\nExample: \"Yes — the validation scenario should be re-run with a 'you have 5 minutes' framing\"\n```\n\n⏸️ **Collect responses.** Store in memory.\n\nIf Question 4 is answered yes, this pipeline's quantitative pass rates don't cover that axis — run `skill-development`'s Phase 3.5 compliance testing (`references\u002Fcompliance-testing.md`) before or alongside this pipeline; see \"Integration with skill-development, skill-refiner-interactive\" in `references\u002Fworkflow.md`.\n\n### Step 2.1b: Cross-Check Against the Target's Own Testing & Validation Section\n\nBefore writing `evals.json`, `Read` the target skill's own SKILL.md for a \"Testing & Validation\" section. If one exists and lists numbered scenarios, check whether the eval scenarios collected in Step 2.1 collectively exercise each item — or explicitly note the gap for any that aren't covered. A skill's own Testing & Validation checklist is a direct, authoritative claim about what \"tested\" should mean for it; writing an eval set that doesn't cross-check against it risks shipping eval coverage narrower than the skill's own documented claims, which reads as tested when it isn't.\n\n### Step 2.2: Generate evals.json\n\nWrite `.\u002Fevals\u002F\u003Cskill-name>\u002Fevals.json` with fields: `skill_name`, `skill_path`, `description`, and `evals` array (each entry: `id`, `name`, `prompt`, `expected_output`, `files`).\n\nSee `references\u002Feval-schema.md` for full schema and annotated examples.\n\n### Step 2.3: Generate eval_metadata.json Files\n\nCreate `workspace\u002Fiteration-1\u002Feval-N\u002Feval_metadata.json` for each eval. Required fields: `eval_id`, `skill_path`, `assertions` array (each entry: `text`, `type`, `target`). Assertion types: `presence`, `quality`, `structure`, `functionality`.\n\nCreate directories: `workspace\u002Fiteration-1\u002Feval-1\u002F`, `eval-2\u002F`, etc. (with_skill and baseline subdirs created in Phase 3).\n\nSee `references\u002Feval-schema.md` for full schema and annotated examples.\n\n### Step 2.4: Plugin-Rule Compliance Assertions (Optional)\n\nWhen evaluating a skill for plugin-rule compliance (naming, tool-scoping, language, formatting), invoke `plugin-rulebook` before writing `eval_metadata.json` to obtain the active rule set. Use each enabled rule as a separate assertion:\n\n```\nSkill: plugin-rulebook\n→ Returns active rules with descriptions, examples, and enforcement levels\n```\n\nAdd rule-compliance assertions to `eval_metadata.json` alongside functional assertions:\n\n```json\n{\"text\": \"allowed-tools uses space-separated format (R6)\", \"type\": \"structure\", \"target\": \"SKILL.md frontmatter\"}\n```\n\n---\n\n## QUICK WORKFLOW (Alternative Path)\n\n**If user chose \"Quick Workflow\" in Step 1.2b, follow this streamlined path instead of Phases 3-7.**\n\n### Quick Phase 1: Run WITH_SKILL Only (No Baseline)\n\n**Goal:** Test the skill itself. No baseline comparison, no timing metrics. Just: Does it work?\n\nFor EACH eval in evals.json, launch ONE agent (no parallel baseline):\n\nSee `references\u002Feval-schema.md` — \"Agent Prompt Templates\" section for the WITH_SKILL_ONLY prompt template (Quick Workflow variant, no timing metrics).\n\n⏸️ **Wait for agent completion before proceeding to next eval.**\n\n### Quick Phase 2: Grade Results (Assertions Only)\n\nFor each eval, grade against assertions in eval_metadata.json. Record pass\u002Ffail + evidence for each assertion.\nSave to: `.\u002Fevals\u002F\u003Cskill-name>\u002Fworkspace\u002Fiteration-1\u002Feval-N\u002Fwith_skill\u002Fgrading.json`\n\nSee `references\u002Feval-schema.md` for grading.json schema.\n\n### Quick Phase 3: Show Results\n\nDisplay simple pass\u002Ffail summary (no benchmark.json):\n\n```\nQUICK VALIDATION RESULTS: \u003Cskill-name>\n====================================\n\nEval 1: \u003CScenario>  ✓ PASS (5\u002F5 assertions)\nEval 2: \u003CScenario>  ✓ PASS (4\u002F5 assertions)\nEval 3: \u003CScenario>  ✗ FAIL (2\u002F5 assertions)\n\nSummary: 11\u002F15 assertions passed (73%)\nStatus: Ready to refine or deploy\n```\n\n### Quick Phase 4: Next Steps\n\nAsk user:\n\n```\nquestion: \"What would you like to do?\"\nheader: \"Next Steps\"\noptions: [\n  {label: \"Run full pipeline\", description: \"Move to complete benchmarking with baseline comparison\"},\n  {label: \"Refine skill\", description: \"Update skill based on failed assertions\"},\n  {label: \"Done\", description: \"Quick validation complete\"}\n]\n```\n\n---\n\n## PHASE 3: Run Tests\n\n**Goal:** Execute 2 agents per eval (WITH skill + BASELINE) in parallel. Capture outputs.\n\n### Step 3.1: Spawn Agents (Parallel Execution)\n\nFor EACH eval, launch 2 agents SIMULTANEOUSLY in one Agent tool call:\n\nSee `references\u002Feval-schema.md` — \"Agent Prompt Templates\" section for the WITH_SKILL and BASELINE prompt templates.\n\n⏸️ **Wait for both agents to complete before proceeding to Phase 4.**\n\n---\n\n## PHASE 4: Grade Results\n\n**Goal:** Evaluate each agent's outputs against assertions in eval_metadata.json.\n\n### Step 4.1: Review Outputs & Grade\n\nFor EACH eval:\n\n1. Read `with_skill\u002Foutputs\u002F` files\n2. Read `baseline\u002Foutputs\u002F` files\n3. For EACH assertion in `eval-N\u002Feval_metadata.json`:\n   - Does with_skill output satisfy the assertion? (true\u002Ffalse)\n   - Does baseline output satisfy the assertion? (true\u002Ffalse)\n   - Collect brief evidence quotes from outputs\n\n### Step 4.2: Write grading.json Files\n\nCreate `workspace\u002Fiteration-1\u002Feval-N\u002Fwith_skill\u002Fgrading.json` and `baseline\u002Fgrading.json`.\nRequired fields: `eval_id`, `configuration` (`\"with_skill\"` or `\"baseline\"`), `assertions_evaluated` array (each: `text`, `passed`, `evidence`), `summary` (`assertions_total`, `assertions_passed`, `pass_rate`).\n\nSee `references\u002Feval-schema.md` for full schema and pass\u002Ffail examples.\n\n---\n\n## PHASE 5: Aggregate\n\n**Goal:** Compute benchmark.json with summary stats (pass rates, tokens, timing).\n\n### Step 5.1: Run Aggregation Script\n\nExecute the aggregation script (see `scripts\u002Faggregate_benchmark.py` for the full implementation; `references\u002Feval-schema.md` for the invocation command and output schema):\n\n```bash\npython ${CLAUDE_SKILL_DIR}\u002Fscripts\u002Faggregate_benchmark.py \\\n  .\u002Fevals\u002F\u003Cskill-name>\u002Fworkspace\u002Fiteration-1\n```\n\nExample:\n```bash\npython ${CLAUDE_SKILL_DIR}\u002Fscripts\u002Faggregate_benchmark.py \\\n  .\u002Fevals\u002Fskill-development\u002Fworkspace\u002Fiteration-1\n```\n\nThis reads all grading.json and timing.json files from the standardized location, outputs `benchmark.json` with per-eval stats (pass rates, tokens, duration, delta) and summary statistics.\n\nSee `references\u002Feval-schema.md` for benchmark.json schema and full example.\n\n---\n\n## PHASE 6: Review Summary\n\n**Goal:** Show user a clear comparison table + next steps.\n\n### Step 6.1: Render Comparison Table\n\nDisplay results in human-readable format:\n\n```\nEVALUATION RESULTS: skill-tester iteration-1\n==============================================\n\nEval 1: [Scenario Name]\n  WITH SKILL:  3\u002F3 (100%) | 2500 tokens | 8s\n  BASELINE:    2\u002F3 (67%)  | 1800 tokens | 5s\n  DELTA:       +33% better | +700 tokens | +3s\n\nEval 2: [Scenario Name]\n  WITH SKILL:  2\u002F3 (67%)  | 2100 tokens | 7s\n  BASELINE:    2\u002F3 (67%)  | 1900 tokens | 4s\n  DELTA:       Even      | +200 tokens | +3s\n\nSUMMARY\n-------\nWith Skill Avg:    96.5% pass rate | 2300 avg tokens\nBaseline Avg:      67.0% pass rate | 1850 avg tokens\nIMPROVEMENT:       +29.5 percentage points\n```\n\n### Step 6.2: Offer Next Steps\n\nAsk user (AskUserQuestion):\n\n```\nquestion: \"What would you like to do?\"\nheader: \"Next Steps\"\noptions: [\n  {\n    label: \"Iterate (update skill, run next iteration)\",\n    description: \"Based on results, modify the skill and run workspace\u002Fiteration-2\"\n  },\n  {\n    label: \"Stop (satisfied with results)\",\n    description: \"Evaluation complete. Save benchmark data for reference\"\n  },\n  {\n    label: \"Refine evals (change test cases, rerun)\",\n    description: \"Update evals.json and run iteration again with same eval set\"\n  }\n]\n```\n\n⏸️ **Collect user input.**\n\n---\n\n## PHASE 7: Iterate\n\n**Goal:** If user chose \"Iterate\", update the skill and run next iteration.\n\n### Step 7.1: Get Improvement Feedback\n\nAsk user (AskUserQuestion, free-form):\n\n```\nquestion: \"What would you like to improve about the skill?\"\nheader: \"Improvement\"\n```\n\n⏸️ **Collect feedback.**\n\n### Step 7.2: Update Skill\n\nUse `skill-refiner-interactive` or direct edits to improve the skill based on feedback:\n- Reword descriptions for clarity\n- Add missing guidance to SKILL.md body\n- Reorganize references\n- Trim token usage\n\n### Step 7.3: Run Next Iteration\n\nCreate `workspace\u002Fiteration-2\u002F` directory. Repeat Phases 3–6 with updated skill.\n\n### Step 7.4: Compare Iterations\n\nShow delta between iteration-1 benchmark and iteration-2 benchmark:\n\n```\nITERATION COMPARISON\n====================\nIteration 1 pass rate: 67%  → Iteration 2: 95% (+28 points)\nIteration 1 tokens:   1900  → Iteration 2: 2100 (+200, acceptable)\n```\n\nLoop back to Phase 6 (Step 6.2) to ask: iterate again or stop?\n\n---\n\n**Suggested next step:** if a benchmark run shows a regression, a failed assertion, or a Quick Workflow FAIL, ask with `AskUserQuestion`: \"Run `enhancement-suggestor` against these results for a classified (complexity\u002Frisk\u002Fbenefit) WHAT\u002FWHY\u002FHOW action plan?\" — options \"Yes\" \u002F \"No\". If yes, invoke the `enhancement-suggestor` agent (via `Agent`) against the results. Never invoke it without asking first.\n\n## Testing & Validation\n\nAfter a test run, verify:\n\n1. **Trigger phrases** — confirm skill activates on: \"run evals on skill-X\", \"benchmark this skill\", \"validate skill performance\", \"compare skill vs. baseline\", \"test skill effectiveness\"\n2. **Non-triggers** — confirm skill does NOT activate on: \"review my skill structure\", \"create a new skill\", \"fix this PR\"\n3. **Mode selection** — Quick Workflow and Full Pipeline branches both produce correctly structured output directories\n4. **Schema integrity** — all JSON files (`evals.json`, `grading.json`, `benchmark.json`, `timing.json`, `eval_metadata.json`) validate against `references\u002Feval-schema.md` schemas\n5. **Baseline parity** — baseline agent receives no SKILL.md content; with_skill agent receives full SKILL.md content\n\n**Verified 2026-07-11:** items 1 and 2 checked directly against the current frontmatter description — each of the 5 trigger phrases maps to specific description language (\"running evals\", \"validating a skill's effectiveness\", \"comparing skill vs. baseline performance\", \"running benchmarks\"), and none of the 3 non-trigger phrases share that vocabulary (they map to `skill-reviewer`'s and `skill-development`'s domains instead). `aggregate_benchmark.py`'s new guard\u002Fsort behavior (items covering script robustness) was verified with synthetic fixtures — see the script's own commit history. A live end-to-end pipeline run (spawning real with_skill\u002Fbaseline agents) has not been performed — that remains the one unverified item.\n\n**Quality gates:**\n- [ ] WITH_SKILL and BASELINE agents launched simultaneously (not sequentially) in Full Pipeline\n- [ ] `benchmark.json` exists in `workspace\u002Fiteration-N\u002F` before presenting Phase 6 summary\n- [ ] All destructive operations (deleting eval dirs) confirmed with user before execution\n\n## Reference Guide\n\n| Resource | Purpose |\n|---|---|\n| `references\u002Feval-schema.md` | JSON schemas, agent prompt templates, and workspace structure |\n| `references\u002Fworkflow.md` | Decision points and detailed workflow guidance |\n| `scripts\u002Faggregate_benchmark.py` | Python script that aggregates grading\u002Ftiming data into benchmark.json |\n| `plugin-rulebook` skill | Active rule configuration for compliance-testing assertions |\n| `enhancement-suggestor` agent | Turns a regression\u002Ffail\u002Fdelta result into a classified, prioritized action plan |\n",{"name":5,"description":7,"allowed-tools":23},"Read Write Edit Glob Grep Agent Bash(python:*) Skill","main","1bf12900cac6f28e346abd2912bdb78d306de723",".agents\u002Fskills\u002Fskill-tester","https:\u002F\u002Fgithub.com\u002FAndreHahm\u002Fandres-cc-marketplace","https:\u002F\u002Fgithub.com\u002FAndreHahm","2026-11-15 04:26:16",{"source":8,"skillName":5,"commands":31,"prompt":41,"zipAvailable":42},[32,35,38],{"tool":33,"command":34},"npx","npx skills add AndreHahm\u002Fandres-cc-marketplace --skill skill-tester",{"tool":36,"command":37},"bunx","bunx skills add AndreHahm\u002Fandres-cc-marketplace --skill skill-tester",{"tool":39,"command":40},"pnpm","pnpm dlx skills add AndreHahm\u002Fandres-cc-marketplace --skill skill-tester","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FAndreHahm\u002Fandres-cc-marketplace\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fskill-tester\u002FSKILL.md",true,[44,50,54,58,62,66,70,76,80],{"path":45,"resourceType":46,"resourceTypeDesc":47,"sizeBytes":48,"sourceHtmlUrl":49},"assets\u002Fbenchmark-example.json","ASSET","资产文件",1414,"https:\u002F\u002Fgithub.com\u002FAndreHahm\u002Fandres-cc-marketplace\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fskill-tester\u002Fassets\u002Fbenchmark-example.json",{"path":51,"resourceType":46,"resourceTypeDesc":47,"sizeBytes":52,"sourceHtmlUrl":53},"assets\u002Fbenchmark-schema.json",1645,"https:\u002F\u002Fgithub.com\u002FAndreHahm\u002Fandres-cc-marketplace\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fskill-tester\u002Fassets\u002Fbenchmark-schema.json",{"path":55,"resourceType":46,"resourceTypeDesc":47,"sizeBytes":56,"sourceHtmlUrl":57},"assets\u002Feval-metadata-example.json",977,"https:\u002F\u002Fgithub.com\u002FAndreHahm\u002Fandres-cc-marketplace\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fskill-tester\u002Fassets\u002Feval-metadata-example.json",{"path":59,"resourceType":46,"resourceTypeDesc":47,"sizeBytes":60,"sourceHtmlUrl":61},"assets\u002Fevals-example.json",1116,"https:\u002F\u002Fgithub.com\u002FAndreHahm\u002Fandres-cc-marketplace\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fskill-tester\u002Fassets\u002Fevals-example.json",{"path":63,"resourceType":46,"resourceTypeDesc":47,"sizeBytes":64,"sourceHtmlUrl":65},"assets\u002Fgrading-baseline-example.json",1069,"https:\u002F\u002Fgithub.com\u002FAndreHahm\u002Fandres-cc-marketplace\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fskill-tester\u002Fassets\u002Fgrading-baseline-example.json",{"path":67,"resourceType":46,"resourceTypeDesc":47,"sizeBytes":68,"sourceHtmlUrl":69},"assets\u002Fgrading-with-skill-example.json",1187,"https:\u002F\u002Fgithub.com\u002FAndreHahm\u002Fandres-cc-marketplace\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fskill-tester\u002Fassets\u002Fgrading-with-skill-example.json",{"path":71,"resourceType":72,"resourceTypeDesc":73,"sizeBytes":74,"sourceHtmlUrl":75},"references\u002Feval-schema.md","REFERENCE","参考文档",9488,"https:\u002F\u002Fgithub.com\u002FAndreHahm\u002Fandres-cc-marketplace\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fskill-tester\u002Freferences\u002Feval-schema.md",{"path":77,"resourceType":72,"resourceTypeDesc":73,"sizeBytes":78,"sourceHtmlUrl":79},"references\u002Fworkflow.md",14294,"https:\u002F\u002Fgithub.com\u002FAndreHahm\u002Fandres-cc-marketplace\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fskill-tester\u002Freferences\u002Fworkflow.md",{"path":81,"resourceType":82,"resourceTypeDesc":83,"sizeBytes":84,"sourceHtmlUrl":85},"scripts\u002Faggregate_benchmark.py","SCRIPT","脚本",9967,"https:\u002F\u002Fgithub.com\u002FAndreHahm\u002Fandres-cc-marketplace\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fskill-tester\u002Fscripts\u002Faggregate_benchmark.py",[87,94,101,108,115,122],{"skillNo":88,"name":89,"nameZh":6,"description":90,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":91,"categories":92,"keywords":93,"publishedAt":18},"01M79R0K918Z0XDM74R0EY5H3P","demo","Fixture skill used by scripts.marketplace_ci sync tests.","https:\u002F\u002Fgithub.com\u002FAndreHahm\u002Fandres-cc-marketplace\u002Fblob\u002Fmain\u002Ftests\u002Fmarketplace_ci\u002Ffixtures\u002Fplugins\u002Fsample-kit\u002Fskills\u002Fdemo\u002FSKILL.md",[],[],{"skillNo":95,"name":96,"nameZh":6,"description":97,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":98,"categories":99,"keywords":100,"publishedAt":18},"01M79R08VMDJA2FPJKH3ZM309K","export-demo","Fixture skill used by scripts.marketplace_ci export tests.","https:\u002F\u002Fgithub.com\u002FAndreHahm\u002Fandres-cc-marketplace\u002Fblob\u002Fmain\u002Ftests\u002Fmarketplace_ci\u002Ffixtures\u002Fclaude\u002Fskills\u002Fexport-demo\u002FSKILL.md",[],[],{"skillNo":102,"name":103,"nameZh":6,"description":104,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":105,"categories":106,"keywords":107,"publishedAt":18},"01M79R042FS581DS2JNW5AZ1MK","work-to-development","Validate an accepted Linear Issue's readiness for implementation (criteria, dependencies, duplicate\u002Fexisting artifacts) and request governed branch\u002Fworktree creation through git-kit:starting-work — recording work-started Git\u002FGitHub evidence and a deliberate Linear Started transition only after Git identity is read back and verified. Use when asked to start work on a Linear issue, begin development for an accepted issue, or prepare an accepted issue for implementation. Never creates a branch directly — always delegates to git-kit.","https:\u002F\u002Fgithub.com\u002FAndreHahm\u002Fandres-cc-marketplace\u002Fblob\u002Fmain\u002Fplugins\u002Fworkmanagement-kit\u002Fskills\u002Fwork-to-development\u002FSKILL.md",[],[],{"skillNo":109,"name":110,"nameZh":6,"description":111,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":112,"categories":113,"keywords":114,"publishedAt":18},"01M79QZZD4EHB82P72P4N92BKT","work-linking","Create and maintain stable cross-system links and authority labels between Notion and Linear records, and detect\u002Frepair drift between them. Use when asked to link a Notion record to a Linear record, check for drift between Notion and Linear, or repair a broken\u002Fstale link. Repairs only the bounded non-authoritative copy or link — never chooses by newest timestamp, and never creates a reverse-write loop against either system's own authority.","https:\u002F\u002Fgithub.com\u002FAndreHahm\u002Fandres-cc-marketplace\u002Fblob\u002Fmain\u002Fplugins\u002Fworkmanagement-kit\u002Fskills\u002Fwork-linking\u002FSKILL.md",[],[],{"skillNo":116,"name":117,"nameZh":6,"description":118,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":119,"categories":120,"keywords":121,"publishedAt":18},"01M79QZVFA3GTVNZMJ51NESVSP","status-and-learning","Produce a dated, explicitly non-live Linear-to-Notion progress summary, or capture an outcome\u002Flearning after work completes. Use when asked to summarize progress to Notion, report portfolio status, or capture what was learned from a completed piece of work — not to disposition open follow-up items (see open-item-management for that). Never runs automatically or continuously — every summary is a deliberate, approved snapshot, not a live sync.","https:\u002F\u002Fgithub.com\u002FAndreHahm\u002Fandres-cc-marketplace\u002Fblob\u002Fmain\u002Fplugins\u002Fworkmanagement-kit\u002Fskills\u002Fstatus-and-learning\u002FSKILL.md",[],[],{"skillNo":123,"name":124,"nameZh":6,"description":125,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":126,"categories":127,"keywords":128,"publishedAt":18},"01M79QZNSKSABK7X2PT93X8QT5","repository-gates","Discover, delegate, record, and invalidate this repository's actual Git\u002FGitHub lifecycle gates (pre-commit, pre-push, PR required checks, Codex delta review, review requirements, merge rights) and the repository-policy provider profile that maps governed operations to git-kit. Use when asked what gates apply to this repository, to discover repository policy before starting Wave 2 work, or to check whether prior gate evidence is still valid after new commits. Discovery and delegation only — never executes a gate's own internal logic, and never invents a gate the repository doesn't actually define (in particular, never assumes a \"Review Changes\" gate exists).","https:\u002F\u002Fgithub.com\u002FAndreHahm\u002Fandres-cc-marketplace\u002Fblob\u002Fmain\u002Fplugins\u002Fworkmanagement-kit\u002Fskills\u002Frepository-gates\u002FSKILL.md",[],[]]