Structured root-cause debugging protocol. Use when a bug has survived one or two fix attempts, when a failure is described as weird, flaky, or impossible, or when you notice you are guessing instead of diagnosing. Also invoked manually via /deep-debug.
Explore-then-plan protocol for complex work. Use before implementing anything that spans multiple files, changes architecture, touches shared interfaces, or has ambiguous requirements. Produces a reviewable plan with acceptance criteria before any code is written. Also invoked manually via /plan-first.
Quality review of your own diff before handing it over. Use after completing any multi-file or non-trivial change, before reporting it done. Catches AI-accent code (noise comments, debug leftovers, scope creep, orphan files) and diff-level bugs. Complements verify-done, which checks that the work RUNS; this checks that the code is CLEAN. Also invoked manually via /self-review.
Evidence check before claiming completion. Use before telling the human a task is done, fixed, or working, especially at the end of long or multi-part tasks. Verifies the work actually runs and every requirement is demonstrably met, then formats an honest report. Complements self-review, which checks code cleanliness; this checks truth. Also invoked manually via /verify-done.