using-agent-skills

Automate task routing by matching incoming tasks to the most suitable agent skill based on metadata, invoking it automatically across session starts and phase-driven engineering flows to reduce manual effort.

Agent skill discovery is the process of matching tasks to suitable skills using metadata fields like name, tagline, and keywords. It detects optional scripts and assets to automatically load and execute the appropriate workflow.

Configure SKILL.md frontmatter by defining required name and description fields, along with optional metadata like complexity, category, keywords, dependencies, and components to enable accurate task-to-skill matching.

You can chain agent skills in lifecycle sequences for phase-based task orchestration, enabling contextual loading of relevant workflows like idea-refine and planning-and-task-breakdown for seamless execution and verification.

Task routing requires a SKILL.md frontmatter with name and description fields. It also detects optional metadata including tagline, request, complexity, category, keywords, dependencies, and components for precise skill matching.

Agent skill discovery works without external dependencies, relying solely on internal SKILL.md frontmatter metadata to match and route engineering tasks to the most suitable available skill automatically.