How the Directory Is Built
The engine scans GitHub for repositories that look like skills, plugins, MCP servers or agents across every supported platform.
Each discovered repository goes through platform detection, quality scoring and AI-assisted categorization into one of the 10 skill types.
The README and the web directory at skills.911fund.io are regenerated from the newly scored set.
Star counts, releases and maintenance status are refreshed daily, with risk enrichment applied where it's still pending.
A weekly summary of new discoveries goes out to the community.
Run schedule#
| Schedule | Stage | Description |
|---|---|---|
| Every 3h | Discovery | Scans GitHub for skills across all platforms |
| Every 3h | Validation | Platform detection, quality scoring & AI categorization |
| Every 3h | Publishing | Regenerates README & web directory |
| Daily 3AM | Metadata Refresh | Updates stars, releases, maintenance status, and pending risk enrichment |
| Weekly Mon 9AM | Discord Digest | Weekly summary of new discoveries |
How platform detection works#
Platform detection is the first validation step — it decides which of the 10 skill types and platform badges a repository earns, before quality scoring runs.
| Icon | Platform | Detection |
|---|---|---|
| 🤖 | Claude Code | SKILL.md, .claude/, plugin.json |
| 🖱️ | Cursor | .cursorrules, .cursor/ |
| 🧠 | Codex | AGENTS.md |
| 🏄 | Windsurf | .windsurfrules |
| 📟 | Cline | .clinerules |
A repository that ships more than one of these files supports more than one platform, and picks up every matching badge.
Next steps#
- How Skills Are Scored — what the scoring step at the center of this pipeline actually measures
- Skill Categories & Types — the 10 categories AI categorization sorts skills into