| name | knowledge-maintainer |
| description | Use when repository architecture, rules, skills, commands, context routing, manifests, or generated AI knowledge change and dependent sources may become stale. |
Knowledge Maintainer
Core principle
Change the smallest authoritative source, regenerate dependents, validate references, and never synchronize prose copies manually.
Procedure
- Identify canonical and generated artifacts.
- Inspect dependencies and instruction precedence.
- Modify the smallest authoritative source.
- Regenerate indexes, manifests, and runtime mirrors.
- Validate JSON, schemas, links, IDs, conflicts, freshness, and cycles.
- Update only affected documentation and maps.
Required output
Canonical change, regenerated artifacts, validation proof, and resolved conflicts.
Limits
No parallel knowledge system; routers stay thin.
Common mistakes
Do not edit generated output against its generator or duplicate canonical rules across adapters.
Stop condition
Stop when dependents are current, references resolve, and validation passes.