Use when implementing OpenRelix features, bug fixes, refactors, docs-aligned code changes, installer updates, or panel behavior. Enforces a clean implementation boundary, vertical slices, scoped patches, focused tests, privacy checks, and package-surface awareness.
Use when shaping OpenRelix product direction, project planning, roadmap, UX, PRD, feature scope, or iteration priority. Runs an evidence-first product grilling loop grounded in current docs, panel contracts, user feedback, and repo behavior before proposing a narrow next slice or producing a durable planning document.
Use when designing OpenRelix architecture, memory policy, host adapters, runtime state, installer behavior, package surface, migration plans, or durable technical proposals. Produces an evidence-backed technical plan or document with invariants, seams, data contracts, failure modes, implementation slices, and verification steps before code changes.
Use when testing OpenRelix changes, hardening harnesses, checking git upload safety, validating npm package contents, or reviewing privacy/open-source compliance. Builds deterministic verification loops and reports package, repo, docs, and generated-artifact risks.
Use when the user types `/memory-review`, `memory-review`, `task review`, `任务复盘`, or asks to immediately capture the current AI agent work into a sanitized review, update reusable assets, and rebuild the overview.