metabolizing-knowledgelisted

# Metabolizing knowledge Every other discipline here produces knowledge. Qualifying a request produces a decision, planning produces a table, a unit produces a verified claim, a scenario produces a validated story. **None of them reclaims any of it**, so the set only grows, and what it grows with is claims that were true once. This skill owns one question: **when does something already marked verified stop being true?** It does not curate decision records, and it does not argue for deleting code. The reason this is a discipline rather than a cleanup is that nothing raises its hand. A requirement whose test was deleted still reads as satisfied. A guard whose implementation was refactored away still has its entry. Left alone the record does not go noisily wrong — it goes quietly wrong, and the day someone relies on it is the day they find out. ## Nothing here works without provenance The handle is a registry linking each claim of completion to the check that verifies it and to the artifact recording that check passing. `traceability.registry`, `traceability.provenance` and `traceability.archive` in `.ledger.yml` name where those live. **A missing adapter is not a reason to do nothing here.** Auditing needs the registry, and a registry can be found: look for it, and if the project keeps one under another name, say which file you are treating as it and why. What genuinely stops on a missing value is **re-verification**, because that has to run this project's checks — there,