What problem does it solve? Moving a skill that matured inside a private host fork (like an OpenClaw deployment) back into the shared gbrain bundle is risky: real names, internal channels, and fork-specific conventions leak easily, and manual copying skips validation. This Skill drives the judgment-heavy editorial workflow so harvested skills are generic, linted, and conformant before they land. ## Core Features & Use Cases - Dry-run preview and privacy lint: Previews the file set with gbrain skillpack harvest --dry-run, then runs a real harvest with a privacy linter that rolls back on any match. - Editorial genericization checklist: A six-point Phase 3 pass that scrubs personal names, real entities, internal channels, fork-specific triggers, and routing-eval examples before copying. - Conformance verification: Validates the result with bun test test/skills-conformance.test.ts, gbrain skillpack check --strict, and a sorted openclaw.plugin.json update. - Use Case: You built a meeting-notes skill in your OpenClaw fork that has run in production for weeks. Use this Skill to scrub your name and company references, harvest it into gbrain, and let every other client scaffold it. ## Quick Start Ask the agent to harvest your mature fork skill into the gbrain bundle, starting with a dry-run preview and the genericization checklist.
