What problem does it solve?
Creating reliable, well-structured autonomous agents for Claude Code plugins is hard: authors must craft precise frontmatter, triggering examples, system prompts, and validate format and behavior to make agents discoverable, safe, and effective.
Core Features & Use Cases
- Agent Templates: Ready-made frontmatter and system prompt patterns for common agent types (reviewers, test generators, docs, security).
- Triggering Guidance: Concrete <example> blocks and commentary patterns to ensure agents activate in the right contexts and avoid false positives.
- Validation & Tooling: Shell-based validation script and reference guides to check frontmatter fields, model/color/tools selection, system prompt structure, and example coverage.
- Use Case: Create a code-review agent for a plugin, validate its frontmatter and examples, then test triggering scenarios before adding it to agents/ to ensure correct discovery and behavior.
Quick Start
Create an agent that reviews pull requests for security and code quality using the agent-creation template and validate it with the provided validation script.