{
  "markdown": "# Claude Skills Registry\n\n> **Core repo:** logic + index + site.  \n> **Main repo (merged artifact):** https://github.com/majiayu000/claude-skill-registry  \n> **Data repo (skills archive):** https://github.com/majiayu000/claude-skill-registry-data  \n> **Authority:** core workflows are canonical; main is a publish mirror.  \n<p align=\"center\">\n  <img src=\"https://img.shields.io/badge/dynamic/json?url=https%3A%2F%2Fmajiayu000.github.io%2Fclaude-skill-registry-core%2Fstats.json&query=%24.archive_skill_md_count_raw&label=SKILL.md%20files%20(raw)&color=blueviolet&style=flat-square\" alt=\"SKILL.md files (raw)\">\n  <img src=\"https://img.shields.io/badge/dynamic/json?url=https%3A%2F%2Fmajiayu000.github.io%2Fclaude-skill-registry-core%2Fstats.json&query=%24.registry_skill_count_dedup&label=Skills%20(dedup)&color=purple&style=flat-square\" alt=\"Skills (dedup)\">\n  <img src=\"https://img.shields.io/badge/dynamic/json?url=https%3A%2F%2Fmajiayu000.github.io%2Fclaude-skill-registry-core%2Fstats.json&query=%24.total_plugins&label=Plugins&color=1f6feb&style=flat-square\" alt=\"Plugins\">\n  <img src=\"https://img.shields.io/badge/dynamic/json?url=https%3A%2F%2Fmajiayu000.github.io%2Fclaude-skill-registry-core%2Fstats.json&query=%24.updated_at&label=Updated%20UTC&color=2ea043&style=flat-square\" alt=\"Updated UTC\">\n  <img src=\"https://img.shields.io/badge/License-MIT-blue.svg?style=flat-square\" alt=\"License\">\n  <a href=\"https://majiayu000.github.io/claude-skill-registry-core/\"><img src=\"https://img.shields.io/badge/Web-Search-blue?style=flat-square\" alt=\"Web Search\"></a>\n</p>\n\n> The most comprehensive Claude Code skills registry — updated daily with the latest skills\n\n## What is this?\n\nThe largest searchable index of Claude Code skills, aggregated from GitHub and community sources.\n\n**Three ways to use:**\n1. **[Web Search](https://majiayu000.github.io/claude-skill-registry-core/)** - Fast browser-based search\n2. **[sk CLI](https://github.com/majiayu000/caude-skill-manager)** - Terminal package manager\n3. **API** - Direct JSON access\n\n**Repo layout note:** `core` owns workflows/pipeline logic, `data` stores `skills/**`, and `main` is generated from `core + data`. See `SCHEME2_SPLIT.md`.\n\n## Generated Mirror Notice\n\nThis repository is the merged publish artifact for browsing and compatibility consumers. Source changes for discovery, download, security scanning, index generation, search, Pages, or publish orchestration belong in [`claude-skill-registry-core`](https://github.com/majiayu000/claude-skill-registry-core). Archived skill body and archive metadata issues belong in [`claude-skill-registry-data`](https://github.com/majiayu000/claude-skill-registry-data).\n\n- **Do not submit normal source PRs here** unless the change is explicitly main-owned, such as publish workflow routing or mirror-only repository metadata.\n- **Release status**: this mirror does not cut independent GitHub Releases for daily generated data. Use `provenance/merge-source.json`, `registry_summary.json`, and the generated timestamps to identify the exact `core_sha` and `data_sha` behind an artifact.\n- **Use issues here only for mirror artifact problems**, such as mismatched generated files, broken compatibility paths, or incorrect provenance after publish.\n\n## Highlights\n\n- **Massive Skill Index** - Deduplicated, high-quality registry (see badge for live count)\n- **Rich Categories** - Development, Testing, DevOps, Design, and more\n- **Daily Updates** - Automated crawling/validation by core scheduled workflows\n- **Quality Indexed** - Metadata, descriptions, and star counts\n- **Lightweight Search** - Gzip-compressed index for fast client-side search\n\n## Operational Ownership\n\n- **Core**: source of truth for workflows, crawling, scanning, and index/site generation\n- **Data**: canonical archived skill tree (`skills/**`)\n- **Main**: publish artifact for merged browsing/compatibility consumers\n- **Publish contract**: core dispatches main publish with pinned `core_sha` + `data_sha`\n- **Non-goal for main**: main does not initiate canonical syncs, archive updates, or index generation\n\n## Quick Start\n\n### Option 1: Web Search\n\nVisit [https://majiayu000.github.io/claude-skill-registry-core/](https://majiayu000.github.io/claude-skill-registry-core/)\n\nFor clone/update tips on large repositories, see [docs/FAST_CLONE.md](docs/FAST_CLONE.md).\n\n### Option 2: CLI (sk)\n\n```bash\n# Install sk\ngo install github.com/majiayu000/caude-skill-manager@latest\n\n# Search skills\nsk search testing\nsk search pdf\nsk search --popular\n\n# Install a skill\nsk install anthropics/skills/skills/docx\n```\n\n### Option 3: Direct API\n\nThe public JSON surface is versioned in\n[docs/artifact-api-contract.md](docs/artifact-api-contract.md). Use that\ncontract when integrating with pointer files, manifests, shards, gzip variants,\nand deprecation markers.\n\n```bash\n# Startup search index and bounded full-search shards\ncurl https://majiayu000.github.io/claude-skill-registry-core/search-index.json\ncurl https://majiayu000.github.io/claude-skill-registry-core/search-index-manifest.json\ncurl https://majiayu000.github.io/claude-skill-registry-core/search-shards/part-000.json\n\n# Deduplicated catalog index with quality/security/install signals\ncurl https://majiayu000.github.io/claude-skill-registry-core/search-index-lite.json\ncurl https://majiayu000.github.io/claude-skill-registry-core/quality-index.json\ncurl https://majiayu000.github.io/claude-skill-registry-core/quality-index-manifest.json\ncurl https://majiayu000.github.io/claude-skill-registry-core/quality-shards/part-000.json\ncurl https://majiayu000.github.io/claude-skill-registry-core/security-index.json\ncurl https://majiayu000.github.io/claude-skill-registry-core/security-index-manifest.json\ncurl https://majiayu000.github.io/claude-skill-registry-core/security-shards/part-000.json\ncurl https://majiayu000.github.io/claude-skill-registry-core/ranking-index.json\ncurl https://majiayu000.github.io/claude-skill-registry-core/ranking-index-manifest.json\ncurl https://majiayu000.github.io/claude-skill-registry-core/ranking-shards/part-000.json\n\n# Lightweight registry summary (counts only)\ncurl https://raw.githubusercontent.com/majiayu000/claude-skill-registry-core/main/registry_summary.json\n\n# Full registry manifest and shards (merged publish artifact)\ncurl https://raw.githubusercontent.com/majiayu000/claude-skill-registry/main/registry-manifest.json\ncurl https://raw.githubusercontent.com/majiayu000/claude-skill-registry/main/registry-shards/00.json\n\n# Compatibility registry pointer (merged publish artifact)\ncurl https://raw.githubusercontent.com/majiayu000/claude-skill-registry/main/registry.json\n\n# Category manifest and bounded parts\ncurl https://majiayu000.github.io/claude-skill-registry-core/categories/index.json\ncurl https://majiayu000.github.io/claude-skill-registry-core/categories/development/manifest.json\ncurl https://majiayu000.github.io/claude-skill-registry-core/categories/development/part-000.json\n\n# Legacy category pointer\ncurl https://majiayu000.github.io/claude-skill-registry-core/categories/development.json\n```\n\n---\n\n## Architecture\n\n```\n┌─────────────────────────────────────────────────────────────────┐\n│  Layer 1: Data Collection                                       │\n│  ┌──────────────┐  ┌──────────────┐  ┌──────────────┐          │\n│  │ GitHub Crawl │→ │ Download     │→ │ Security     │          │\n│  │ (discover)   │  │ (sync)       │  │ (scanner)    │          │\n│  └──────────────┘  └──────────────┘  └──────────────┘          │\n└─────────────────────────────────────────────────────────────────┘\n                              ↓\n┌─────────────────────────────────────────────────────────────────┐\n│  Layer 2: Index Generation                                      │\n│  ┌────────────────┐  ┌────────────────┐  ┌────────────────┐    │\n│  │ search-index   │  │ categories/    │  │ featured.json  │    │\n│  │ .json          │  │ *.json         │  │ (featured set) │    │\n│  └────────────────┘  └────────────────┘  └────────────────┘    │\n└─────────────────────────────────────────────────────────────────┘\n                              ↓\n┌─────────────────────────────────────────────────────────────────┐\n│  Layer 3: Consumption                                           │\n│  ┌────────────────┐  ┌────────────────┐  ┌────────────────┐    │\n│  │ Web UI         │  │ sk CLI         │  │ API            │    │\n│  │ (GitHub Pages) │  │ (Go)           │  │ (JSON)         │    │\n│  └────────────────┘  └────────────────┘  └────────────────┘    │\n└─────────────────────────────────────────────────────────────────┘\n```\n\n### Search Index Schema\n\n```typescript\n// Lightweight index for fast search\ninterface SearchIndex {\n  v: string;           // Version (date)\n  t: number;           // Indexed skill count (always s.length)\n  s: SkillMini[];      // Skills array\n}\n\ninterface SkillMini {\n  n: string;           // name\n  d: string;           // description (truncated 80 chars)\n  c: string;           // category code (dev, ops, sec, etc.)\n  g: string[];         // tags (max 5)\n  r: number;           // stars\n  i: string;           // install path\n}\n```\n\n---\n\n## Directory Structure (Core)\n\n```\nclaude-skill-registry-core/\n├── registry.json           # Lightweight core registry metadata\n├── docs/                   # GitHub Pages\n│   ├── index.html          # Web search UI\n│   ├── search-index-lite.json\n│   ├── search-index.json   # Compatibility pointer to bounded shards\n│   ├── search-index-manifest.json\n│   ├── search-shards/      # Bounded full search index parts\n│   ├── featured.json       # Featured skills snapshot\n│   └── categories/         # Category manifests and bounded parts\n├── sources/                # Data sources\n│   ├── anthropic.json\n│   ├── community.json\n│   └── skillsmp.json\n├── scripts/                # Build scripts\n│   ├── build_search_index.py\n│   ├── discover_by_topic.py\n│   ├── security_scanner.py\n│   └── ...\n└── (no committed skills/)  # skills/** lives in registry-data; mounted in CI when needed\n```\n\nThe merged `claude-skill-registry` publish artifact additionally contains\n`registry-manifest.json` and `registry-shards/**`.\n\n---\n\n## Categories\n\nCanonical category slugs, short codes, governance status, and heuristic keywords\nlive in `taxonomy/categories.yaml`. Pipeline scripts read that file instead of\nkeeping their own category lists. Legacy category names are diagnostic inputs\nonly; they are routed to review and are not accepted as publish categories.\nThe canonical category proposal is documented in\n`docs/plan/canonical-taxonomy-proposal.md`, and the current target count report\nis `docs/plan/canonical-category-target-counts.json`.\nValidate the taxonomy itself with `python scripts/check_taxonomy_governance.py`.\nTo validate publish output against canonical categories, pass repeated\n`--publish-category <slug>` values. To review category quality across the\narchive before moving anything, run\n`python scripts/audit_category_quality.py --skills-dir skills`.\nThe default audit uses metadata and paths for a fast full-archive pass; add\n`--include-frontmatter` when checking frontmatter/category drift.\nThe audit also reports non-standard nested skill paths such as\n`category/category/skill/SKILL.md`.\nFor those layout issues, use\n`python scripts/normalize_skill_depth.py --skills-dir skills --json` to review\nthe exact move plan before applying it.\nFor semantic reclassification, generate a review-only migration plan with\n`python scripts/plan_category_migration.py --skills-dir skills --output category-migration-plan.json`.\nThe plan records action, confidence, source categories, target category, keyword\nsignals, and reason for every proposed change; it does not move files.\nFor a bounded second-pass model review, set `MIMO_API_KEY` and run\n`python scripts/review_category_plan_with_llm.py --plan category-migration-plan.json --output category-llm-review.json --checkpoint-jsonl category-llm-review.checkpoint.jsonl --resume`.\nThe default endpoint is `https://token-plan-sgp.xiaomimimo.com/v1` with\n`mimo-v2.5-pro`. The default request uses `--thinking disabled` and\n`--max-completion-tokens 1024` so the model output budget is reserved for the\nrequired JSON; raise the token value for larger reviews, or pass\n`--thinking default` when using a non-MiMo endpoint that does not accept the\nprovider-specific thinking field. The report remains review-only: it records\nmodel category, confidence, decision, parse status, token/thinking policy, and\nevidence without modifying the archive.\nThe checkpoint file is append-only JSONL, so interrupted long reviews can resume\nwithout re-calling the model for completed candidates.\n\nCategory counts are published in `categories/index.json`; full category payloads\nare available through `categories/<category>/manifest.json` and bounded\n`part-*.json` files. The legacy `categories/<category>.json` URL is now a small\ncompatibility pointer. The publish path runs\n`python scripts/check_category_artifacts.py --categories-dir docs/categories` so\nlegacy category files cannot silently grow back into large full-payload JSON.\nCommon category codes include:\n\n| Category | Code | Description |\n|----------|------|-------------|\n| `development` | `dev` | Development tools, frameworks |\n| `data` | `dat` | Data processing, analysis |\n| `design` | `des` | UI/UX design, frontend |\n| `testing` | `tst` | Testing, QA, automation |\n| `devops` | `ops` | DevOps, CI/CD, infrastructure |\n| `documents` | `doc` | Document creation (docx, pdf) |\n| `productivity` | `pro` | Productivity and automation |\n| `product` | `prd` | Product management |\n| `security` | `sec` | Security, auditing |\n| `marketing` | `mkt` | Marketing, content, SEO |\n\n---\n\n## Roadmap\n\n### Current Status\n\n- [x] **Index count** tracked by the badge (core `registry.json`)\n- [x] **Archive size:** tracked by badge (raw `SKILL.md` count from `stats.json`)\n- [x] **Daily auto-update** via GitHub Actions\n- [x] **Security scanning** for all skills\n- [x] **sk CLI** for installation\n\n### In Progress\n\n- [x] **Lightweight search index** (gzip-compressed; see stats.json)\n- [x] **Web search UI** (GitHub Pages)\n- [x] **GitHub Pages deployment** (https://majiayu000.github.io/claude-skill-registry-core/)\n\n### Planned\n\n- [ ] **AI semantic search** (vector similarity)\n- [ ] **Skill recommendations** (based on usage)\n- [ ] **Version tracking** for skills\n- [ ] **Skill quality scoring**\n- [ ] **API rate limiting** and caching\n\n---\n\n## Contributing\n\n### Add Your Skill\n\n**Option 1: Submit via Issue**\n1. Open an [issue](https://github.com/majiayu000/claude-skill-registry-core/issues/new)\n2. Use the \"Add Skill\" template\n3. Provide: repo URL, name, description, category\n\n**Option 2: Submit via PR**\n1. Fork `majiayu000/claude-skill-registry-core`\n2. Add your skill to `sources/community.json`:\n\n```json\n{\n  \"name\": \"your-skill-name\",\n  \"repo\": \"your-username/your-repo\",\n  \"path\": \"optional/path/to/skill\",\n  \"description\": \"What your skill does\",\n  \"category\": \"development\",\n  \"tags\": [\"testing\", \"automation\"]\n}\n```\n\n3. Submit a PR to `majiayu000/claude-skill-registry-core` (not `claude-skill-registry`, which is a generated publish artifact)\n\n### Report Issues\n\nWe welcome feedback! Please open an issue for:\n- **Bugs** - Search not working, incorrect data\n- **Feature requests** - New categories, better search\n- **UX improvements** - Web UI, CLI enhancements\n- **Data quality** - Duplicate skills, wrong categories\n\n👉 [Open an Issue](https://github.com/majiayu000/claude-skill-registry-core/issues/new)\n\n### Contribute Code\n\n```bash\n# Clone the core repo (authoritative pipeline repo)\ngit clone --filter=blob:none --sparse https://github.com/majiayu000/claude-skill-registry-core.git\ncd claude-skill-registry-core\n\n# Pull only what you need (add more paths later as needed)\ngit sparse-checkout set --cone docs scripts sources schema\n\n# Install dependencies\nuv venv\nsource .venv/bin/activate\nuv pip install -r requirements.txt\n\n# Build search index locally\npython scripts/build_search_index.py --registry registry.json --output docs\n\n# Test web UI\ncd docs && python -m http.server 8000\n# Visit http://localhost:8000\n```\n\nSee `docs/FAST_CLONE.md` for more options (existing clones, getting full checkout, Windows notes).\n\n---\n\n## The Agent Infra Stack\n\nThis project is one layer of an open-source stack for running coding agents (Claude Code, Codex) as serious infrastructure. Every piece works standalone; together they close the loop:\n\nIn this stack, the registry is the **Extend** layer — the front door where you discover skills. Scan third-party skills with `argus` before installing, and keep agents honest at runtime with `vibeguard`.\n\n| Layer | Project | What it does |\n|---|---|---|\n| Extend | [claude-skill-registry](https://github.com/majiayu000/claude-skill-registry) **◀ you are here** | Discover and search community Claude Code skills |\n| Extend | [spellbook](https://github.com/majiayu000/spellbook) | Cross-runtime skills for Claude Code, Codex, and multi-agent workflows |\n| Trust | [argus](https://github.com/majiayu000/argus) | Static install-time scanner for supply-chain attacks (npm / PyPI / crates.io) |\n| Trust | [vibeguard](https://github.com/majiayu000/vibeguard) | Rules, hooks, and guards against hallucinated or unverified agent changes |\n| Remember | [remem](https://github.com/majiayu000/remem) | Local-first persistent memory for Claude Code and Codex sessions |\n| Orchestrate | [harness](https://github.com/majiayu000/harness) | Rust agent orchestration platform — rules, skills, GC, observability |\n| Route | [litellm-rs](https://github.com/majiayu000/litellm-rs) | High-performance Rust AI gateway — 100+ LLM APIs via OpenAI format |\n| Keep | [keepline](https://github.com/majiayu000/keepline) | Session command center — monitor, recover, never lose agent work |\n\n---\n\n## Related Projects\n\n| Project | Description |\n|---------|-------------|\n| [caude-skill-manager](https://github.com/majiayu000/caude-skill-manager) | CLI tool for installing skills (`sk`) |\n| [anthropics/skills](https://github.com/anthropics/skills) | Official Anthropic skills |\n| [SkillsMP](https://skillsmp.com) | Web-based skill marketplace |\n| [awesome-claude-skills](https://github.com/travisvn/awesome-claude-skills) | Curated skill list |\n\n---\n\n## License\n\nMIT License applies to the registry code/pipeline only - see [LICENSE](LICENSE) for details.\n\n## Third-Party License & Attribution\n\nThird-party skills under `skills/**` keep their original licenses and copyright ownership.\n\n- Repository-level MIT does **not** relicense third-party skill content.\n- Every imported skill metadata file should include:\n  - `author`\n  - `source_url`\n  - `license`\n  - `copyright`\n  - `permission_note`\n  - `distribution` (`compatible` or `restricted`)\n- `restricted` entries are not MIT-compatible by default and require explicit upstream permission before redistribution/use.\n- Notices are generated by compliance checks into `THIRD_PARTY_NOTICES.md`.\n- Metadata compliance runs in advisory mode by default to avoid blocking ingestion; strict blocking can be enabled when needed.\n\n---\n\n<p align=\"center\">\n  Made with ❤️ for the Claude Code community\n</p>\n",
  "bytes": 19182,
  "sha": "8f2ce99cd4c66197a8a256c006cc6c05473f74178b899dd48962a3bcb921b779",
  "repo_slug": "majiayu000/claude-skill-registry",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/skl_majiayu000_claude_skill_registry_skills__23c1b3f0/readme"
}