YaoMetaSkill：工程评测系统为

YAO = Yielding AI Outcomes. A rigorous 工程, 评测, governance,、portability 系统为 reusable agent skills.GitHub 1,847 ⭐ / 175 Fork，主要使用 Python 开发。标签：agent-skills,ai-agents,evaluation,governance,meta-skill。访问项目：https://github.com/yaojingang/yao-meta-skill

项目介绍

YAO = Yielding AI Outcomes. A rigorous 工程, 评测, governance,、portability 系统为 reusable agent skills.

核心功能

- Skill IR:平台-neutral intermediate representation 为 intent, triggers, inputs, outputs, boundaries, references,、expected artifacts. - Target compilers、adapters: generated surfaces 为 OpenAI, Claude, generic agent skills, Agent Skills compatible packages,、VS 代码-oriented workflows. - Output Eval 实验室: 触发 checks, output assertions, execution evidence, timing、令牌 evidence, 基准测试 reproducibility, blind-审查 packs, answer keys,、adjudication reports. - 审查 工作室 2.0:single HTML gate 页面为 intent, triggers, output eval, 上下文 cost, 运行时 checks, trust, Skill Atlas signals, adoption drift, waivers, annotations, 发布 evidence, warnings, blockers,、fix actions. - Evidence、发布 governance: evidence consistency checks, 包 verification, install simulation, 运行时 permission probes, world-类 evidence intake, world-类 ledger, operator runbook,、public claim guard. - SkillOps loop: metadata-only adoption drift, telemetry hooks, adaptive proposals, daily、weekly curator reports,、portfolio-级别 drift signals.

使用指南

`yao-meta-skill` 创建, evaluates, packages,、governs reusable agent skills. The 1.0 line focused 在 turning repeated workflows 进入 installable, readable, 跨平台 skill packages. The 2.0 line expands factory 进入 a Skill OS:governed 系统为 modelingskill once, compiling it 为 multiple targets, 测试它的 behavior, reviewing 它的发布 evidence,、trackingnext iteration.

项目信息

项目地址

GitHub 仓库：<https://github.com/yaojingang/yao-meta-skill>

> 完整 URL：[https://github.com/yaojingang/yao-meta-skill](https://github.com/yaojingang/yao-meta-skill)

如需查看完整源码、安装命令、使用示例、API 文档，请直接访问上方项目地址。所有信息均以 GitHub 仓库为准。