What problem does it solve? It automates the discovery, evaluation, and traceable absorption of external skills from sources like GitHub Topics, Hermes skill directories, and arXiv, preventing untracked or unsafe ad-hoc adoption of third-party capabilities. ## Core Features & Use Cases - Multi-Source Scanning: Rotates keyword-based scans across GitHub Topics (every cycle), Hermes skills/ directories (every cycle), and arXiv (every two cycles). - Unified Ledger Tracking: Records every discovered project in absorption-ledger.json with a strict state lifecycle: tracking → evaluating → absorbed/deferred/archived. - Five-Layer Absorption Records: Requires a detailed evolution/absorption-{project}.md record per deep absorption, plus safety checks that reject unverified executable code. - Use Case: During an evolution cycle, scan GitHub Topics for agent-skills repositories, evaluate a candidate project, write its five-layer absorption record, and append its state transition to the central ledger. ## Quick Start Scan external skill sources using the keyword rotation strategy and record any newly discovered projects into absorption-ledger.json with their initial tracking state.
