What problem does it solve? Designing a Claude Code plugin requires deciding how many agents to create, which skills belong to which movement, which hooks are compatible with which lifecycle events, and how commands should route. Doing this ad hoc leads to orphaned skills, incompatible hook configurations, and unmaintainable plugin structures. ## Core Features & Use Cases - Agent Role Definition: Assigns skills to agents using a 5+ skills heuristic, with boundaries, least-privilege tool allowlists, and maxTurns estimates. - Hook Compatibility Validation: Checks proposed hooks against the type-event compatibility matrix (command/http work on all 22 events; prompt/agent only on ToolUseContext events) and flags CRITICAL mismatches. - Command Routing & Quality Gates: Builds a canonical command table with aliases and places testable quality gates between movements. - Use Case: After running an ideation step that produced a plugin brief, feed the brief path to this Skill to produce an architecture-plan.md with Mermaid diagrams, a component summary, and next-step recommendations. ## Quick Start Run the plan-architecture skill with the path to my plugin brief to generate a complete architecture plan document.
