Knowledge infrastructure that documents your codebase automatically, for agents and humans
Why consider it
- Inverts documentation economics by capturing rather than requiring writing
- Ties doc updates to the change that caused them, which removes drift structurally
- Consolidating context sources resolves duplication and contradiction between systems
- The built-in library index addresses agents getting third-party APIs wrong
Worth knowing
Trade-offs: Captured wrongness is worse than no documentation, since it will be trusted · Whether captured knowledge is reviewable before becoming authoritative is unstated