[FEATURE] Skill discovery command — browse and install community skills from a registry
Preflight Checklist
- [x] I have searched existing requests and this feature hasn't been requested yet
- [x] This is a single feature request (not multiple features)
Problem Statement
Claude Code skills are powerful but hard to discover. Right now, finding useful skills requires:
- Knowing they exist
- Searching GitHub manually
- Cloning repos and copying files into
~/.claude/skills/
There's no central way to browse, search, or install community-created skills. I maintain awesome-claude-skills as a curated list, but a static markdown file isn't a real discovery mechanism.
The Agent Skills spec exists, but there's no built-in way to leverage it for discovery.
Proposed Solution
Add a /skills command (or extend the existing skill system) with:
/skills search "data visualization" # Search community skills
/skills browse # Browse by category
/skills install <github-url> # Install from git URL
/skills list # List installed skills
/skills update # Update installed skills
This could pull from:
- A community registry (like npm for skills)
- GitHub topics (e.g.,
claude-code-skill) - The Agent Skills spec metadata
Feature Area
Configuration and settings
Use Case Example
- I'm starting a new data analysis project
- I run
/skills search "pandas matplotlib" - Claude shows me 5 matching skills with descriptions and install counts
- I pick one:
/skills install mingrath/data-viz-skill - It clones to
~/.claude/skills/data-viz-skill/and is immediately available - Next session, I can run
/skills updateto get the latest version
Additional Context
The skills ecosystem is growing fast but fragmented. A discovery mechanism would:
- Lower the barrier for new users to benefit from community skills
- Incentivize skill authors to follow the Agent Skills spec
- Create a network effect that makes Claude Code more valuable
Reference: awesome-claude-skills — a curated list of 150+ skills I maintain, which shows the demand for discovery.
This issue has 2 comments on GitHub. Read the full discussion on GitHub ↗