knowledge base hygiene

Knowledge-base hygiene is the discipline of catching and containing the small structural and content failures that compound across a personal knowledge system — a misquoted citation propagates into every downstream retrieval, a skeleton card with a generic parent masquerades as a real reference, and a metadata-only book sits unverified in the citation chain for months. Effective hygiene separates concerns rather than collapsing them: quote-mangling is a presentation bug fixed at the surface, internal provenance blocks are sanitized at the visitor boundary rather than censored at the source, and the agent's generation loop must slow down enough for the diagnostic taxonomy to be applied to every new entry. The underlying principle is asymmetry of cost — a single unverified quote, a missing callout, or a bare wikilink is cheap to fix in isolation but expensive once downstream pages, retrieval pipelines, and public mirrors inherit the error. Hygiene therefore treats the knowledge base as a layered system where the source of truth stays rich and messy, but every surface — retrieval, publication, verification — operates against a cleaned, checked version.