Summary
This is meta-tooling for multi-agent systems: it crawls your repository looking for skill.yaml files and builds a capability map so a master agent can figure out which sub-agent to route tasks to. Think of it as service discovery but for AI skills. You'd run this at startup or when an agent needs to decide whether to invoke the HIPAA checker or the PII sanitizer for a given request. It only reads metadata, doesn't execute anything, and will skip malformed YAML with a warning. Honest take: if you're building a single-purpose agent, you don't need this. But if you're orchestrating multiple specialized skills and need programmatic routing, it solves a real coordination problem.
Install to Claude Code
npx -y skills add jorgealves/agent_skills --skill agent-capability-discovery --agent claude-codeInstalls into .claude/skills of the current project.
Keep your Mac awake while Claude Code and 40+ AI agents run. Sleeps when they're idle.
One time payment $9 →
Integrate web data into your AI product. One API to scrape website & brand data.
Get API Key Now →
Agent, run crypto. Access onchain data & trade routes via 1inch.
Install now →
On Capafy, your Skill runs online 24/7 as an agent product, and you get paid every time someone uses it.
Start earning →
Files
SKILL.mdView on GitHub
Agent Capability Discovery
Purpose and Intent
The agent-capability-discovery skill is the "brain" of a multi-agent system. It allows an agent to self-reflect and understand what tools and expertises are available within its environment by indexing all skill.yaml files.
When to Use
- System Initialization: Run this when an agent starts up to populate its internal tool list.
- Routing Decisions: Use this when a master agent receives a complex user request and needs to identify which specialized skill (e.g.,
hipaa-compliance-guardvs.pii-sanitizer) is best suited for the job. - Documentation Generation: Automatically keep a "Global Skills Map" up to date for human developers.
When NOT to Use
- Skill Execution: This skill only discovers what is possible; it does not execute the other skills.
- External Tooling: It only indexes skills defined in the local repository structure.
Input and Output Examples
Input
base_directory: "."
output_format: "markdown"
Output
A markdown table or JSON object listing all discovered skills, their descriptions, and their primary capabilities.
Error Conditions and Edge Cases
- Broken YAML: If a
skill.yamlis malformed, the discovery tool will skip that directory and report a warning. - Circular Dependencies: This tool does not handle execution dependencies, only discovery.
Security and Data-Handling Considerations
- Metadata Only: The tool only reads the definition files; it never touches actual source code or data unless instructed by the indexed skills themselves.
- Local Scope: Discovery is confined to the provided root directory to prevent directory traversal attacks.
Featured
Keep your Mac awake while Claude Code and 40+ AI agents run. Sleeps when they're idle.
One time payment $9 →Integrate web data into your AI product. One API to scrape website & brand data.
Get API Key Now →Agent, run crypto. Access onchain data & trade routes via 1inch.
Install now →On Capafy, your Skill runs online 24/7 as an agent product, and you get paid every time someone uses it.
Start earning →
