What problem does it solve? When you hand an AI a minimal or seemingly inaccessible source (a URL that won't load, an image, a bare name, or a text fragment) and want it turned into concrete improvements for your skill ecosystem, most workflows stop at "unavailable" or apply changes blindly. This Skill maximizes information extraction from any source and runs the full harvest pipeline with real machine validation. ## Core Features & Use Cases - Information Maximization (acquire front-end): Graceful fallback ladder for any source type — direct fetch, search for fetchable mirrors, domain analysis, and trend validation — with an honest Source-Reality record of what was obtained versus proxied. - Thin orchestration without duplication: Delegates deep research to rlm-harness, backlog and execution to skill-ecosystem-auditor, validation to validation-mesh, and state persistence to continuation-memory. - Real machine validation: Applies approved backlog items only after re-reading full file content, snapshotting, and diffing, then verifies with skill_guard.py and audit_scan.py instead of grep. - Use Case: You paste a Telegram channel link that cannot be fetched directly and say "research this for updating my skills" — the Skill falls back through search and domain analysis, maps real trends against your ecosystem, and applies only validated, non-duplicate improvements. ## Quick Start Ask the AI to research this source for updating your skills, then paste any URL, image, name, or text fragment you want harvested.
