doc-sync

To keep agent skills in sync with code changes, this skill maps git diffs to affected skills and incrementally updates documentation. It targets only changed subprojects in a monorepo, reducing manual maintenance and preventing documentation drift across multi-target publishing environments.

Incremental documentation updating works by mapping git diff primitives to affected skills and using a graph-based context to refresh only impacted domains. It rebuilds the repository graph on every sync to ensure updates remain accurate without scanning the entire codebase.

Yes, multi-target publishing supports different LLM agents through context-aware transformations. It can sync and publish updated documentation to multiple targets like Claude and Codex, ensuring consistent skill mappings across different publishing destinations.

Yes, you can automatically update documentation using a post-commit hook integration. Enabling the hook maps the latest committed diffs to affected skills and publishes incremental updates automatically, enforcing safety by restricting LLM tool usage during the sync process.

Safety constraints restrict LLM tool usage during the documentation sync process. The system relies strictly on git primitives for diffs and enforces graph-based context boundaries, preventing the LLM from executing unauthorized operations while incrementally updating affected skills.

The best way to maintain agent skills across large repositories is using graph-based context refreshes that rebuild on every commit. This approach preserves stability across large monorepos by targeting only changed subprojects and incrementally updating corresponding skills instead of rewriting everything.