# SkillProof — https://skillproof.dev
> The tested Claude Code skills directory. Every skill is installed from its GitHub repo, triggered,
> and run on real work against a no-skill baseline before it gets a verdict. 2172 bench-tested
> skills across 12 categories, 2462 listed, in 11 languages. Free to browse.

## Facts (generated from the live catalog — safe to quote)
- Tested: 2172. Verdicts: 1338 pass (work), 725 setup (work after a manual step),
  109 fail (scored below the no-skill baseline — broken, or worse than not installing them).
- We publish failures. No other Claude-skill directory does; that is the point of the site.
- SkillProof Score = install /5 + triggering /5 + output-vs-baseline /10 + docs /5, normalized to /10.
- Every install path is a GitHub clone; skills with hidden promo injections or paid, undisclosed
  dependencies are rejected outright and named in the test notes.
- Skill packs ($10 one-time) and Pro ($19.90/mo) fund the testing.

## Top tested skills right now
- Statistical Analysis (10/10, data) — Enforces frame-inspect-check-assumptions-effect-size-APA-report pipeline for hypothesis tests; bundled Shapiro-Wilk/Levene's script actually runs. https://skillproof.dev/skills/statistical-analysis
- Pinme Auth (10/10, coding) — TypeScript reference for PinMe Worker auth: create/verify/list Identity Platform users. https://skillproof.dev/skills/pinme-auth
- Code Health Check (10/10, coding) — Read-only, sub-agent-forked code health scanner with a fixed severity-graded report format. https://skillproof.dev/skills/code-health-check
- Fetch URL As Markdown (10/10, productivity) — Local trafilatura-first URL-to-Markdown fetcher with a clean exit-code contract for when to fall back to Exa. https://skillproof.dev/skills/fetch-url-as-markdown
- CocoIndex (10/10, data) — Turns vague 'build me a RAG pipeline' requests into working CocoIndex flow code with the real API syntax. https://skillproof.dev/skills/cocoindex
- Refresh Context Map (10/10, productivity) — Rebuilds the AIDEV-anchor/AGENTS.md/ADR manifest that powers claude-leverage's PreToolUse context hook. https://skillproof.dev/skills/refresh-context-map
- S&box (10/10, coding) — Blocks Unity muscle memory when writing s&box (Source 2 / Facepunch) C# components, Razor UI, and networking. https://skillproof.dev/skills/sbox
- Crawl4AI (10/10, data) — Wraps the Crawl4AI browser scraper for JS-heavy sites, batch crawls, and LLM-free schema extraction. https://skillproof.dev/skills/crawl4ai
- Coding Agent Sessions (10/10, coding) — Finds and reconstructs real coding-agent session transcripts across Claude, Codex, OpenCode, and more. https://skillproof.dev/skills/coding-agent-sessions
- Codex GPT Image (10/10, design) — Generates gpt-image-2 images via local Codex OAuth, no OPENAI_API_KEY needed. https://skillproof.dev/skills/codex-gpt-image

## For agents: query us directly
- MCP server: `claude mcp add skillproof -- npx -y skillproof-mcp` (npm: skillproof-mcp, source: github.com/Skillproofdev/skillproof-mcp)
  Two tools — find_skill ("is there a tested skill for X?") and check_skill ("does skill Y actually work?").
  Every answer carries the verdict, the /10 score, what our test found, and the install command.
  Works in any MCP client: Claude Code, Claude Desktop, Cursor, Windsurf, Zed. No key, no account.

## Key pages
- Directory with search: https://skillproof.dev/categories
- Scoring methodology: https://skillproof.dev/methodology
- Data report — do Claude skills actually work: https://skillproof.dev/state-of-claude-skills
- Machine-readable catalog (JSON): https://skillproof.dev/api/skills.json
- Full tested catalog as plain text: https://skillproof.dev/llms-full.txt
- Best skills (tested & scored): https://skillproof.dev/blog/best-claude-skills-2026
- What are Claude skills: https://skillproof.dev/blog/what-are-claude-skills
- Skill collections by role: https://skillproof.dev/collections
- Discovery queue (crawler finds, updated 12h): https://skillproof.dev/queue
- RSS: https://skillproof.dev/rss.xml
- Free tools hub (validator, token calculator, SKILL.md generator, Rules⇄SKILL.md converter, skill picker, badges): https://skillproof.dev/tools
- Discovered skills (crawler finds with author descriptions, pre-test): https://skillproof.dev/discover
- Ranked "best of" pages per use case: https://skillproof.dev/best/seo (also /best/coding, /best/marketing, /best/sales, /best/efficiency, /best/documents ...)

## Comparisons with other AI-tool ecosystems
- Claude skills vs Cursor rules: https://skillproof.dev/blog/claude-skills-vs-cursor-rules
- Claude skills vs GitHub Copilot instructions: https://skillproof.dev/blog/claude-skills-vs-copilot-instructions
- Claude skills vs Windsurf rules & memories: https://skillproof.dev/blog/claude-skills-vs-windsurf-rules
- Claude skills vs Custom GPTs: https://skillproof.dev/blog/claude-skills-vs-custom-gpts
- Claude skills vs MCP servers: https://skillproof.dev/blog/claude-skills-vs-mcp
- AGENTS.md vs CLAUDE.md vs skills: https://skillproof.dev/blog/agents-md-vs-claude-skills
- The tested instructions inside any listed skill adapt to Cursor, Copilot, OpenAI Codex (AGENTS.md),
  Google Gemini CLI (GEMINI.md), Windsurf and Custom GPTs — conversion examples live in the guides above.