Use when implementing OpenSpec tasks, writing code, running tests, or fixing bugs. Guards TDD flow, execution modes, and spec compliance.
Use when creating, reviewing, or updating OpenSpec proposals, designs, tasks, spec deltas, acceptance criteria, or archiving changes.
Use when diagnosing and fixing bugs, applying security patches, or making small enhancements with /team-repair. Guards DIVE protocol, dual-track closure, and evidence-driven repair.
Use when exploring requirements, doing deep design exploration, or brainstorming before proposing. Guards OpenSpec environment and brainstorm output location.
Use when reviewing task breakdowns, checking dependency ordering, optimizing task granularity, or deciding whether to proceed to implementation.
Use when implementing an approved OpenSpec change, progressing through tasks, updating task status, or continuing an incomplete implementation.
Use when retrospecting on AI collaboration, recording bad cases, updating skill rules, or managing skill versions.
Use when doing pre-merge code review, checking requirement alignment, security risks, test coverage, and regression risk.