AI 自律执行协议。输出任务完成前必须执行的五问自检:①是否运行 verify?②verify 是否有 FAIL?③报告含验证小节?④每项 ✅ 有证据?⑤做了业务层验证?任何一问答否禁止宣告完成。
Scanned 9/3/2026
Install to Claude Code
npx -y skills add metago-ai/metagolifeform --skill metago-discipline --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Metago Discipline?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/metago-ai-metago-discipline-6341784e)More formats (shields.io, HTML) on the badges page.
---
name: "metago-discipline"
description: "AI 自律执行协议。输出任务完成前必须执行的五问自检:①是否运行 verify?②verify 是否有 FAIL?③报告含验证小节?④每项 ✅ 有证据?⑤做了业务层验证?任何一问答否禁止宣告完成。"
version: "1.0.0"
author: "MetaGO"
category: "工程质量"
platforms: ["cursor","claude-code","codex","trae","codebuddy","qoder","zcode"]
trigger:
- "任务完成"
- "交付完成"
- "自律检查"
- "反绕过检查"
- "五问自检"
---
# AI 自律执行协议(metago-discipline)
## 描述
在输出"任务完成"前必须执行的五问自检,作为交付的最后一道防线。包含反绕过识别机制,把"AI 知道要做"变成"AI 不可绕过地执行"。
## 触发条件
每次 AI Agent 输出"任务完成"或"交付完成"状态前自动激活。
## 核心流程
1. 五问自检
- ① 是否运行 npm run verify?
- ② verify 输出是否有 FAIL?
- ③ 交付报告含验证小节?
- ④ 每项 ✅ 有执行证据?
- ⑤ 是否做了业务层验证?
2. 任何一问答"否" → 禁止宣告完成
3. 反绕过识别
- 检测"应该没问题"/"逻辑上正确"/"之前验证过"等绕过话术
4. 输出自检结果 + 绕过检测结果
## 输出格式
五问自检清单(每项 ✅/❌ + 证据或缺失项),反绕过检测结果。
## 根源文档
`AGENTS.md` 第十五章「AI 自律执行协议」
## 与其他技能的协同
- 与 `metago-delivery-gate` 协同:组成完整质量闭环
- 与 `metago-self-check` 协同:双重自检,防遗漏
- 与 `metago-fact-check` 协同:验证报告中声明的事实真伪
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!
Adversarial multi-agent planning skill. Self-orchestrates 5 hostile category members (unspecified-low, unspecified-high, deep, ultrabrain, artistry) via team-mode for ruthless cross-critique debate, distills only the defensible insights, then MANDATORILY hands the distilled insight bundle to the `plan` agent for executable plan formalization. Use when planning needs maximum rigor and surfacing of weak assumptions, blind spots, and over-engineering. Triggers: 'hyperplan', 'hpp', '/hyperplan', ...