Use ONLY when the user explicitly requests thinking-methodology work: 理清混乱思路 / 表达, 问题定义与根因分析, 方案设计与决策, 创新破局, 复盘与思维沉淀, 论证 / 推理 / 方案 Review, 思维质量验收, or mentions 元逻辑, 元法则, 子思维, 思维体系, meta-logic, meta-logic-method. Loads the meta-logic-method thinking system and dispatches tasks per the AGENTS.md 调度表 to
Meta-Logic Thinking System (meta-logic-method)
Three-layer thinking rule system: meta-logic laws (meta/) > sub-thinking skills (sub-skills/) > reference (reference/), plus workflow compositions (workflows/).
Entry point
AGENTS.md is the single rule entry. Read it before any thinking task; it contains the task-to-document dispatch table and the output protocols.
How to use
- Classify the task via the AGENTS.md 调度表, then load ONLY the documents it names (do not load the whole tree).
- Respect the four-layer constraint: 元法 > 强制/禁止 > 建议 > 可选; resolve conflicts at the stricter level.
- Output analyses / decisions using the AGENTS.md §4 思维链输出协议(问题 → 概念 → 前提 → 推理 → 反证 → 结论+置信度 → 行动+重审条件).
- Review tasks use the AGENTS.md §5 审查输出协议([Blocking] / [Non-blocking] / [Info],引用 M-x / S-x / W-x / F-x / B-x 编号).
- Before finishing an analysis, self-check the README §六 思维质量验收清单(11 items).
Key files
| Purpose | File |
|---|---|
| AI dispatch rules & output protocols | AGENTS.md |
| Ten meta laws M-1~M-10 + M-T / M-P / M-L details | meta/ |
| 12 sub-thinking skills S-1~S-12 | sub-skills/ |
| 5 workflows W-1~W-5 | workflows/ |
| Fallacies F-x / biases B-x / CoT templates R-x / glossary | reference/ |
| Numbered rule index (script-generated, do not edit) | INDEX.md |
Maintenance: meta/ is human-review-only (AI may propose changes via the 体系建议:... format); see GOVERNANCE.md for versioning, numbering, and change permissions.
Related skills
- Downloads
- 195All time
- Security audit
- Pass
- Last updated
- 1w ago
- Current version
- v0.1.2
- License
- MIT-0
