Use when questioning why safety gates exist, debating whether a mechanical gate applies, or constructing justifications to route around constraints. Provides meta-awareness context that complements CLAUDE.md Layer 0 (mechanical gates) and Layer 1 (behavioral guidelines).
Use when adding/updating dependencies, running installs, debugging package issues, reviewing build output for network activity, or evaluating package safety. Defines supply-chain hygiene, registry verification, AI contribution disclosure, and anomaly reporting rules that complement CLAUDE.md Layer 0/1.
Use when reference docs (API-REFERENCE, SCHEMA-REFERENCE, MIGRATIONS-LOG) are stale, after adding endpoints/models/migrations, before releases, or when issue asks for doc regeneration