# Meta Recursive Discovery

> Enabling agents to autonomously search for and adopt new specialized skills.

- Skill: `jcorpac/meta-recursive-discovery` (Agent Skill)
- Install (CLI): `npx skillmds@latest add jcorpac/meta-recursive-discovery`
- Raw SKILL.md: https://api.skillmd.com/api/skills/jcorpac/meta-recursive-discovery/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Product & Planning
- Author: jcorpac (https://skillmd.com/u/jcorpac)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/jcorpac/meta-recursive-discovery

---


# Meta-Recursive Discovery

An agent should know what it doesn't know and where to find the answers.

## The Discovery Loop
1.  **Impediment Detection**: Identifying when a task requires a skill not currently in the active context.
2.  **Library Search**: Using search tools to look for specialized skills in the root directory.
3.  **Self-Correction**: Updating the current approach based on the found skill's instructions.

## Autonomous Improvement
- **Template Usage**: Using the `template-skill` to create *new* temporary skills when a unique problem is solved, effectively growing the library's wisdom over time.

## Best Practices
- **Documentation Hygiene**: Maintaining a clean index so discovery remains fast.
- **Verification**: Always cross-reference discovered skills with established best practices (e.g., ADRs).