yao-meta-skill

Why we included this project

Anyone who has rewritten the same prompt for ChatGPT, Claude, and a VS Code extension will recognize the problem this project tackles: it packages repeated workflows, prompts, runbooks, and transcripts into installable agent skills that work across tools. The key move is modeling a skill once in a neutral intermediate representation, then compiling it for OpenAI, Claude, generic agent skills, and VS Code-oriented targets, so the same capability moves between assistants without being re-typed. Around that packaging sits a governance layer that most skill tools skip: an output evaluation lab with assertions and execution evidence, a review studio that gates releases, package verification and install simulation, and telemetry that shows whether a skill is still being used after release. That makes it a fit for a lead or small team that wants to turn a personal prompt into a maintained team asset with release evidence, or keep a portfolio of skills from quietly rotting. It is a framework and methodology rather than a ready-made agent, so expect to invest in the workflow rather than plug it in and go.