What problem does it solve? Teams working with AI agents need a single, authoritative place to record binding behavioural directives, but ad-hoc rule writing leads to duplicates, vague unenforceable instructions, and rules files polluted with ephemeral notes. ## Core Features & Use Cases - Duplicate-checked rule creation: Reads the existing rules file and checks every ### Rule: heading before appending, surfacing overlaps and suggesting amendments instead of duplicates. - Schema-validated entries: Enforces a three-part structure (Rule title, Directive, Rationale) via a JSON Schema, with a validate-rules.sh script that can validate the file or generate correctly formatted entries. - Scope discipline: Rejects vague directives and ephemeral notes, routing memory-style requests to vault-capture instead of the rules file. - Use Case: A user says "New rule: never merge a PR touching the delivery Lambda without running the retry-sweep acceptance test." The skill reads the existing rules, generates a schema-valid entry with an ALWAYS/NEVER directive and rationale, validates the file, and confirms the addition. ## Quick Start Ask the agent to add a new rule, for example: "New rule: always run the typecheck before committing changes to the i18n dictionary."
