القائمة
Four-layer cognitive distillation system. Use when performing daily reviews, recording decisions, extracting patterns, or updating principles. Transforms raw data into durable wisdom through systematic compression: Raw → Episodic → Semantic → Principles.
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
استنادا إلى تصنيف SOC المهني
Writes comprehensive READMEs serving both human readers and AI agent readers. Use when user says "用 GitHub Skill", "productize README", or asks to create a product-grade README for a repo.
Use when you have a written implementation plan to execute in a separate session with review checkpoints
Use when receiving code review feedback, before implementing suggestions, especially if feedback seems unclear or technically questionable - requires technical rigor and verification, not performative agreement or blind implementation
Use when completing tasks, implementing major features, or before merging to verify work meets requirements
Create new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy.
Use when implementing any feature or bugfix, before writing implementation code
| name | cognitive-distillation |
| description | Four-layer cognitive distillation system. Use when performing daily reviews, recording decisions, extracting patterns, or updating principles. Transforms raw data into durable wisdom through systematic compression: Raw → Episodic → Semantic → Principles. |
| platform | both |
| scope | agent:wendy |
A four-layer system that transforms raw information into durable principles through progressive compression. Each layer distills the one below it.
L4: PRINCIPLES.md ← 原子原则 (每周 review)
L3: memory/patterns.md ← 认知模式 (每日蒸馏)
L2: memory/decisions/ ← 决策日志 (实时记录)
L1: memory/*.md + Obsidian ← 原始数据 (自动积累)
Files: memory/YYYY-MM-DD.md, Obsidian briefings, conversation logs
Update: Continuous / automatic
Purpose: Capture everything. Don't filter at this stage.
Already handled by OpenClaw's memory flush + cron briefings.
Files: memory/decisions/YYYY-MM-DD-<topic>.md
Update: Whenever a significant decision is made
Purpose: Record WHY, not just WHAT.
Decision log template:
# Decision: <title>
Date: YYYY-MM-DD
Context: <what situation triggered this decision>
Options:
1. <option A> — pros / cons
2. <option B> — pros / cons
3. <option C> — pros / cons
Selected: <which option>
Reason: <why this one>
Expected outcome: <what we expect to happen>
Actual outcome: <filled in later>
Lesson: <filled in later>
When to create a decision log:
File: memory/patterns.md
Update: Daily distillation cron (22:00)
Purpose: Extract recurring patterns from L1 + L2.
Format:
# Patterns
## Park's Working Style
- <observation> (first noted: YYYY-MM-DD, confirmed: N times)
## Market Patterns
- <pattern> (evidence: ...)
## System/Engineering Patterns
- <pattern> (evidence: ...)
## Anti-Patterns (things that failed)
- <what> → <why it failed> (date)
Rules:
File: PRINCIPLES.md (workspace root)
Update: Weekly review (Sunday 20:00)
Purpose: The most compressed, highest-value knowledge.
Format:
# Principles
## Investing
- **<principle>** — <one-line explanation> [weight: 1-10] [last validated: date]
## Building
- **<principle>** — <one-line explanation> [weight: 1-10] [last validated: date]
## Meta / Thinking
- **<principle>** — <one-line explanation> [weight: 1-10] [last validated: date]
Rules:
memory/patterns.md if warrantedQuick scan: is there an unlogged significant decision from recent conversation? If yes: create L2 decision log. If no: pass.
Cron jobs needed:
0 22 * * * → isolated agentTurn0 20 * * 0 → isolated agentTurnFiles to create on first run:
mkdir -p memory/decisions
touch memory/patterns.md
touch PRINCIPLES.md