agent-capability-discovery_skill

Overview

This skill scans a repository of agent skill directories to build a comprehensive global map of available agent capabilities, inputs, and outputs. It indexes skill definition files to produce a searchable catalog that a master agent or developer can use to route requests or generate documentation. The result is a single source of truth describing what each sub-agent can do.

How this skill works

The skill locates and parses skill definition files (e.g., skill.yaml) under a specified base directory, extracting metadata such as name, description, inputs, outputs, and tags. It validates and skips malformed files while reporting warnings, then aggregates findings into a structured output (JSON, Markdown, or table) suitable for programmatic consumption or human review. The tool never executes skills or read arbitrary source code; it only reads declared metadata within the repository scope.

When to use it

• On system startup to populate an agent’s internal tool registry and available capabilities.
• When a master agent must choose the best specialized skill to handle a complex user request.
• To generate or update a Global Skills Map for developer documentation or audits.
• Before routing tasks to ensure required inputs and outputs match a sub-agent’s contract.
• When onboarding new skills so the catalog stays current and discoverable.

Best practices

• Keep skill definition files consistent and well-structured to improve discovery accuracy.
• Run discovery automatically on CI or at startup to ensure the registry reflects the latest skills.
• Treat discovery results as read-only metadata; use separate runtime processes for execution.
• Limit the base directory to intended repo paths to avoid accidental indexing of unrelated files.
• Log and review warnings for malformed definitions to maintain a healthy skill ecosystem.

Example use cases

• A master agent examines the catalog to pick between a privacy sanitizer and a compliance checker for a data-handling task.
• Generate a Markdown table of all skills for internal documentation and developer onboarding.
• Verify which skills declare a required input type before composing a multi-step workflow.
• Detect missing or duplicate capability declarations during a code review or CI run.
• Produce a JSON payload of capabilities for another service that performs routing or scheduling.