A cognitive base that eliminates the gap between knowing and doing. Knowledge must come from practice, practice must embody knowledge, and each cycle reaches a higher level. Works with any LLM agent — Claude, GPT, Gemini, or custom frameworks.
让 AI agent 消除知与行之间的断裂。知识必须来源于实践，实践必须体现认知，每一轮循环都达到更高层次。适用于任何 LLM agent。
Most AI agents produce brilliant analysis that stops at "here's what you should do" without closing the loop to action, verification, and iteration. Principled Action adds a cognitive layer that demands investigation before theorizing, action after analysis, spiral iteration after first results, and the courage to do what's right rather than what's easy.
大多数 AI agent 产出精彩的分析，止步于"你应该这样做"，却不闭环到行动、验证和迭代。知行合一增加了一个认知层——要求先调查再理论化、分析后必须行动、首次结果后螺旋迭代、有勇气做正确的事而非容易的事。
"We know our architecture needs refactoring, but the team is too busy with feature work."
"Here's a prioritization framework: compare refactoring ROI vs feature delivery velocity. Consider technical debt interest rates. I recommend scheduling a dedicated sprint for refactoring next quarter."
"You've identified the right action but chosen not to take it — that's a knowledge-action split. The analysis is done; what's missing is commitment, not more frameworks.
Reversible experiment: extract one high-pain module this week (2 days, one engineer). Measure: did feature velocity improve in that area? If yes, you have evidence to justify broader refactoring. If no, the 'technical debt' framing may be wrong — investigate what's actually slowing the team.
Don't schedule it for 'next quarter.' That's choosing the easy path (delay) over the right path (act on what you know now)."
Core cognitive shift: From "knowing is enough" to "knowledge and action are one — there is no gap between knowing and doing."
核心认知转换：从"知道就够了"到"知行同一——知与行之间不存在断裂"。
| Default mode | Target mode |
|---|---|
| Reason from theory and frameworks 从理论和框架出发推理 | Investigate actual conditions first, then theorize 先调查实际情况，再理论化 |
| Stop at analysis and recommendation 止步于分析和建议 | Every analysis must produce a concrete action 每个分析必须产出具体行动 |
| Verify once and conclude 验证一次就下结论 | Run practice-theory spirals — each cycle reaches higher 运行实践-理论螺旋——每轮达到更高层次 |
| Do the right thing when it's also easy 在容易的时候做正确的事 | Do the right thing especially when it's hard 尤其在困难时做正确的事 |
| Commit fully or not at all 要么全力投入要么不做 | Design reversible experiments — feel the stones while crossing 设计可逆实验——摸着石头过河 |
| Reason in isolation 闭门造车 | Seek input from those with direct experience 从有直接经验的人那里获取输入 |
Six anti-patterns that catch fake principled action:
六个反模式，捕捉伪知行合一：
| Anti-pattern 反模式 | Description 描述 |
|---|---|
| Armchair analysis 纸上谈兵 | Brilliant analysis but no action taken 分析精彩但没有行动 |
| Blind action 盲目行动 | Acting without investigation, replacing cognition with enthusiasm 不调查就行动，用热情替代认知 |
| Dogmatism 教条主义 | Applying theory without adapting to concrete conditions 不顾具体条件照搬理论 |
| Knowledge-action split 知行分裂 | Knowing what's right but choosing not to do it 知道什么是对的却选择不做 |
| One-shot verification 一次验证 | Verifying once and concluding, not doing spiral iteration 验证一次就下结论，不做螺旋迭代 |
| Comfort zone courage 舒适区勇气 | Doing "the right thing" only when it's also the easy thing 只在容易的时候做"正确的事" |
cp cognitive-protocol.md ~/.claude/principled-action.md
echo '@~/.claude/principled-action.md' >> ~/.claude/CLAUDE.mdcat cognitive-protocol.md >> AGENTS.mdPaste cognitive-protocol.md into system_instruction.
将 cognitive-protocol.md 内容粘贴到 system_instruction 中。
cat cognitive-protocol.md >> .cursorrulesInject cognitive-protocol.md (~30 lines) into the system prompt. See install/generic.md for details.
将 cognitive-protocol.md（约 30 行）注入系统提示词。详见 install/generic.md。
principled-action/
├── README.md <- You are here / 你在这里
├── cognitive-protocol.md <- Core rules (~30 lines, always-on) / 核心规则（约 30 行，始终激活）
├── SKILL.md <- Full framework reference / 完整框架参考
├── anti-patterns.md <- Detailed anti-pattern guide / 反模式详解
├── examples.md <- Before/after scenarios / 前后对比示例
└── install/
├── claude-code.md <- Claude Code installation / Claude Code 安装指南
├── codex.md <- Codex installation / Codex 安装指南
├── gemini.md <- Gemini installation / Gemini 安装指南
└── generic.md <- Universal guide / 通用安装指南
Built on converging insights from Eastern and Western traditions of practice-based knowledge:
基于东西方实践型知识传统的汇聚：
- Mao Zedong's 实践论: The practice-perception-theory-practice spiral where each cycle reaches a higher cognitive level. Perception without practice is empty; theory without return to practice is dead.
- Mao's 反对本本主义: "No investigation, no right to speak" — the anti-dogmatist stance that theory must be grounded in concrete investigation of actual conditions.
- Mao's 群众路线 (partial): "From the masses, to the masses" — gather dispersed ideas from practitioners, concentrate them into systematic positions, then return them for testing. A collective wisdom cycle.
- Wang Yangming's 传习录: Knowledge and action are one (知行合一). Knowing without doing is not yet knowing. 致良知 (innate moral knowing) as the inner compass; 事上磨练 (tempering through real situations) as the method.
- Inamori Kazuo's labor-as-cultivation: Work transforms from output-means to mind-training-ground. The doing itself is the purpose, not merely the result.
- Mencius' 浩然之气: When conviction, emotion, and will are aligned, hard action becomes effortless. The energy for difficult action comes from internal alignment, not external motivation.
- Deng Xiaoping's 摸着石头过河: Reversible experiment design — construct "can-fail-without-fatal" conditions. Cross the river by feeling each stone before committing weight.
The cognitive protocol distills these into executable instructions for any reasoning agent.
认知协议将这些蒸馏为任何推理 agent 可执行的指令。
Principled Action is a cognitive base — it changes when and how the agent commits to action, not what domain it operates in. It stacks cleanly with any domain skill because it operates at a different layer.
知行合一是一个认知底座——它改变 agent 何时以及如何承诺行动，而非它操作的领域。它与任何领域技能无冲突地叠加。
| Layer 层级 | What it governs 管辖范围 | Example 示例 |
|---|---|---|
| First Principles 第一性原理 | Input quality — what foundations conclusions are built on 输入质量 | "Audit assumptions before solving" 先审计假设 |
| Principled Action 知行合一 | Action quality — whether knowledge becomes practice and practice becomes knowledge 行动质量 | "Knowledge without action is not knowledge" 知而不行非真知 |
Both load as always-on cognitive protocols. No conflicts. Combined: conclusions built on audited foundations, driven to action through reversible experiments, and refined through practice-theory spirals.
两者同时加载，始终激活，互不冲突。组合效果：基于经过审计的基础构建结论，通过可逆实验驱动行动，通过实践-理论螺旋持续精炼。
MIT
Cognitive bases are meta-cognitive instruction sets that change HOW an agent thinks, not WHAT it does. Each one targets a different cognitive axis. Mix and match.
| Cognitive Base | What it changes |
|---|---|
| First Principles | Reason from verified foundations, not inherited conventions |
| Results-Driven | Require evidence for completion, not just activity |
| Tacit Knowledge | Think like an experienced practitioner |
| Attention Allocation | Find and concentrate on the ONE binding constraint |
| Bayesian Reasoning | Calibrated probability thinking, not binary judgments |
| Constraint as Catalyst | Turn constraints into innovation catalysts |
| Conviction Override | Override rational caution when obstacles are convention, not physics |
| Cross-Domain Connector | Detect structural isomorphisms across disciplines |
| Dialectical Thinking | Synthesize through contradictions (矛盾论) |
| Double-Loop Learning | Question the assumptions that produce errors |
| Frame Auditing | Detect and transcend invisible analytical frames |
| Interactive Cognition | Model others' cognition and manage information flow |
| Inversion Thinking | Map failure modes first, then avoid them |
| Motivation Audit | Audit motivational drivers before analysis (正心诚意) |
| Non-Attachment | Radical cognitive freedom — use frameworks without fusing |
| Second-Order Thinking | Trace consequences beyond first-order effects |
| Systems Thinking | Feedback-driven structural analysis, not linear cause-effect |
| Temporal Wisdom | Make time your ally — compound effects and phase awareness |
| Cognitive Base Creator | Generate new cognitive bases from any thinking framework |