About OpenSkillIndex

Skills for agentic AI systems are scattered across GitHub repos, marketplaces, blog posts, and Twitter threads. OpenSkillIndex is a neutral registry that indexes them all — harness-agnostic, free to use, open to contributors.

What we index

What makes this neutral

OpenSkillIndex is not affiliated with Anthropic, OpenAI, OpenClaw, or any single harness vendor. Every skill page shows its source, its ecosystem tag with our confidence level, and the authoritative install link that sends you back to the original repo or marketplace. We don't re-host code. We don't sell placement.

How ecosystem detection works

Each skill is matched against a rule matrix with explicit confidence levels (heuristic / strong / verified). When the signal is ambiguous, the ecosystem is marked unknown rather than forced into a bucket. Authors can claim their skills (via GitHub OAuth, coming soon) and override.

Ranking

Search uses Postgres full-text rank plus trigram name similarity, with an exact-match short-circuit for slug and canonical name, log-scaled popularity, and a freshness penalty for skills that haven't been confirmed working recently. The full formula is versioned in the open-source ranking module.

Open source

Frontend + API + schema + migrations + ranking formula + ecosystem detection rules + reference GitHub crawler are open source under the MIT license. Production crawler schedules, anti-abuse heuristics, secrets, and exact enrichment prompts are kept private for operational-security and abuse-prevention reasons — not as a moat.

GitHub org: openskillindex (public repositories launching with the v1.0 release).

Takedowns & contact

If your skill should not be indexed, email support@openskillindex.com and we'll hide it within 24 hours. See /bot for crawler policy.

Credits

Data sourced from public GitHub repositories, surfaced with attribution. Design built with Next.js 15, Tailwind, Postgres+pgvector. Typography: Fraunces (serif), Inter (sans), JetBrains Mono.