What problem does it solve? After running pattern analysis on multiple repositories, learnings sit scattered across per-repo files with no way to see which patterns recur universally, which conflict, and which are worth generalizing across codebases. ## Core Features & Use Cases - Cross-Repo Synthesis: Reads all patterns.md learnings files from ~/.claude/repo-learnings/ and identifies universal patterns, convergent solutions, complementary pairs, and conflicting trade-offs. - Meta-Principle Distillation: Derives 4-8 overarching engineering principles with concrete examples and consequences of violation. - Persistent Output: Writes the full synthesis to ~/.claude/cross-repo-learnings/synthesis.md with an index for ongoing reference. - Use Case: After analyzing five OSS repos with /analyze-patterns, run this Skill to discover that compound components and thin handlers appear everywhere, while mock isolation strategies genuinely conflict — and get guidance on when each applies. ## Quick Start Ask the AI to synthesize patterns across all repos, optionally filtering by repo name or pattern area such as testing.