What problem does it solve? Over time a project's captured memory store accumulates duplicates, misplaced facts, stale status claims, and noisy hooks. This Skill performs a periodic batch consolidation pass over the current project's knowledge tree so memory shrinks and sharpens instead of growing unbounded. ## Core Features & Use Cases - Tree-wide dedup and merge: Scans every level under the anchor (including sibling projects) with a controlled duplicate scanner that plants a paraphrase control to prove the instrument fired, then merges near-duplicates at their common parent. - Placement and re-leveling: Routes each fact through a routing prompt against scope descriptors, moving facts up or down the tree with a corroboration gate for tree-top promotion. - Integrity verification and pruning: Backs up the store with an order-independent manifest, reconciles pointer indexes, sweeps for status rot and voice issues, and prunes obsolete entries content-first. - Use Case: When a SessionStart nudge reports a memory consolidation is due, run the dream to capture pending learnings, deduplicate across sibling projects, re-level misplaced facts, and finish with a verified clean manifest diff. ## Quick Start Ask the AI to run a dream consolidation on the current project's memory tree, capturing this session's learnings first and then deduplicating, re-leveling, and pruning the store.
