What problem does it solve? Repository documentation drifts as code evolves: facts get duplicated across files, public-surface changes ship without reference updates, and tutorials, runbooks, and architecture notes get mixed into the wrong places. This Skill manages documentation as a dependency-ordered system of canonical artefacts so every fact lives in exactly one location and stays aligned with implemented behaviour. ## Core Features & Use Cases - Nine ordered documentation steps: Covers foundation entrypoints, architecture and ADRs, reference, explanation, how-to guides, tutorials, operations runbooks, audience indexes, and governance/lifecycle files like CHANGELOG.md and SECURITY.md. - Four operating modes: establish creates missing structure, sync updates docs after code changes, and audit/pre-pr-review report drift findings without modifying files. - Drift detection: Flags missing, misplaced, duplicated, conflicting, or stale documentation using change-to-doc heuristics tied to code, contracts, and tests. - Use Case: After changing a CLI's flags and configuration, run the skill in sync mode with the reference step to update docs/reference/ first, then audit the diff before opening a pull request to catch any undocumented user-visible changes. ## Quick Start Ask the AI to run the system-documentation skill in audit mode across the whole repository to report documentation drift by class and canonical target file.
