Yao Meta Skill
A meta-skill that creates, evaluates, packages and governs reusable agent skills
- 2,732 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- Runs Python scripts for evaluation, packaging and reports
- The self-update feature changes the installation after explicit confirmation
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/yao-meta-skillDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Runs Python scripts for evaluation, packaging and reports
- The self-update feature changes the installation after explicit confirmation
This is third-party code. Review the repository files before installing.
What it does
The skill turns repeated workflows, prompts and scripts into reusable skill packages and tracks their lifecycle. At its core is a platform-neutral intermediate representation of a skill that captures intent, triggers, inputs, outputs, boundaries and expected artifacts, and compilers generate variants for several targets, including OpenAI, Claude, the generic Agent Skills format and a VS Code oriented environment. A separate evaluation layer covers trigger checks, output assertions, execution evidence, benchmark reproducibility and blind-review packs. Release governance adds evidence consistency checks, install simulation, a runtime permission probe and a single HTML review page. There are modes from a rough draft to a strict governed release and a post-publication loop that watches for drift.
Who it is for. For developers and skill authors who want engineering discipline in creating, evaluating and shipping reusable skills.
Good fit when
- You need to turn a stable workflow or prompt set into a reusable skill
- You need to improve or rebuild an existing skill and add trigger and output evals
- You need a pre-release check: packaging, install simulation and evidence
Not a fit when
- You need a one-off prompt, a summary, a translation or docs rather than a reusable skill
- There is no repeated scenario and stable output contract worth packaging into a skill
Example request
Create a skill from this workflow and add trigger evalsLimitations
The skill targets engineering work on skills and relies on Python scripts in the scripts folder and a Makefile target. Some judge-backed evals and cross-provider output checks need access to model APIs: the Anthropic and OpenAI APIs are unavailable from Russia without a VPN, so those features may not work. The author honestly marks some strong world-class claims as evidence-gated: provider-backed, blind human and telemetry evidence are tracked as separate tasks rather than completed work. The daily update check contacts GitHub.
How to disable. Remove the yao-meta-skill folder from the agent's skills directory, for example from the Codex or other agent install where you placed it. The installed update channel stops working together with the folder.
FAQ
How does 2.0 differ from 1.0?
The 1.0 line turned repeated processes into installable skills. The 2.0 line expands this into a Skill OS: a semantic contract, target compilers, an output eval lab, a review page and release and drift governance.
Does the skill invent metrics?
No; it is built into the rules: it marks unavailable telemetry, approvals, metrics and benchmarks as missing evidence and does not fabricate evidence.
Discussion
No questions or reviews yet.
Related
SkillsA meta-skill that turns a workflow into an agent skill: prior-art search, trigger evaluation and publication to GitHub
A skill that designs, tests, evaluates and packages other agent skills, starting from architecture
A meta-skill that takes a vague idea to a finished agent skill: it designs, writes the files, checks them and refines by evidence
A skill for creating skills that proves them by behavior: before-and-after runs, regression evals and ecosystem audit
A skill that helps an agent author other skills to the agentskills.io spec: structure, metadata, progressive disclosure and validation
Meta Skill (ms)
Meta Skill
Meta Skill (ms) keeps skills locally in SQLite and Git, searches them with a hybrid index and serves agents over MCP
README in short
The README explains that YAO stands for Yielding AI Outcomes and that the goal is not more prompt text but reusable assets and real outcomes. The 2.0 line expands the skill factory into a Skill OS: an intermediate representation, target compilers, evaluation evidence, release gates and operation reports. Install goes through the skills CLI, first globally for Codex and then optionally for every agent. After installing you restart the client and ask to create, improve or evaluate a skill. There is a daily update check and a single confirmation before updating an install. Docs are available in English, Chinese, Japanese, French and Russian. MIT licensed.
README badge
Original SKILL.md
--- name: yao-meta-skill description: Create, improve, or evaluate an existing skill from workflows, prompts, SOPs, scripts. Use for migration/release/package, routing, evals/tests, install/trust checks, 优化已有 skill, 补 trigger 评测. Exclude summary/translation/docs, brainstorming, one-off prompts, copy-only edits, outline-only requests. metadata: author: Yao Team --- # Yao Meta Skill ## Router Rules - Route by frontmatter `description` using the lightest mode. - Keep `SKILL.md` lean: guidance in `references/`, logic in `scripts/`, evidence in `reports/`. ## Modes - `Scaffold`: exploratory; `Production`: team; `Library`: shared; `Governed`: release-critical. - Rules: [Method](references/skill-engineering-method.md), [Modes](references/operating-modes.md), [Boundaries](references/resource-boundaries.md). ## Compact Workflow 1. Require `repeated use` + `reusable output contract`; skip one-offs. 2. Lock target path + identity before writes; self-edits require `--self`. 3. Capture job, output, exclusions, constraints, standards, mode.