What problem does it solve? Agents working with a personal knowledge base need to understand how bases are structured, which queue to read, when pages expire, and which narrower kb skill handles a given job — without this map they misfile captures, drain the wrong queue, or run destructive commands against the wrong base. ## Core Features & Use Cases - Base layout orientation: Documents the anatomy of a base (.kb/, _raw/, wiki zones, index.md) and the three tests for pending/, work/, and cache/. - Queue and query guidance: Distinguishes kb inbox (agent-scoped ingest work) from kb pending list --where waits_on=human, and covers kb find/kb search with quoted date comparisons. - Safe lifecycle operations: Prescribes dry-run-first kb prune and reason-required kb archive, always with an explicit --base, plus wikilink conventions and an authority model for reads, report-only findings, and destructive actions. - Use Case: A user asks "why wasn't my note kept?" or "run base maintenance" — the agent uses this skill to inspect pending entries, dry-run a prune on the named base, and route the actual filing to kb-capture or kb-recall as appropriate. ## Quick Start Ask the agent to explain how your knowledge base is organized and to dry-run a prune on a named base to show which pages would expire.