What problem does it solve? Session observations and bugfix notes accumulate in agent memory (MEMORY.md) without ever becoming durable, searchable lessons, causing knowledge loss and memory bloat across sprints. ## Core Features & Use Cases - Gap Detection: Mines recent bugfix and decision observations and compares them against existing project lessons in docs/lessons/ and cross-project patterns in 00_meta/patterns/. - Lesson Authoring & Pattern Promotion: Writes new lessons with standard frontmatter, registers them in the lessons index, and promotes recurring lessons to cross-project pattern files. - Memory Maintenance: Trims MEMORY.md sections duplicated in CLAUDE.md, updates the Last Crystallized date, and stamps the vault via the dotf CLI. - Use Case: After finishing a major sprint, run the ritual to convert two weeks of unvaulted bugfix observations into indexed lessons and shrink an overgrown MEMORY.md. ## Quick Start Ask the agent to run the crystallize ritual to vault recent observations and clean up MEMORY.md.