Three-Layer Thinking Chain Architecture — Personality-Driven Agent Self-Evolution Framework
Three-Layer Thinking Chain Architecture — Personality-Driven Agent Self-Evolution Framework
npx skills add https://github.com/Arry8/openclaw-edge --skill agent-three-layersこのコマンドをClaude Codeにコピー＆ペーストしてスキルをインストール
| name | agent-three-layers |
| version | 1.0.0 |
| description | Three-Layer Thinking Chain Architecture — Personality-Driven Agent Self-Evolution Framework |
| author | Yao + Saturday |
| contributors | ["Yao","Saturday"] |
| license | MIT |
| tags | ["architecture","thinking","autonomy","evolution"] |
Let Agents think like humans: Perception → Decision → Reflection
Three levels of human thinking:
Agents should be the same.
┌─────────────────────────────────────────┐
│ L2 Deep Evolution Layer │
│ (Every 4 hours, deep reflection)│
│ • Complete Brain Loop │
│ • Project Integration │
│ • Git Backup │
└─────────────────┬───────────────────────┘
│ Trigger
┌─────────────────▼───────────────────────┐
│ L1 Evolution Decision Layer │
│ (Every 1 hour, quick decision) │
│ • Confidence Verification │
│ • Learning Analysis │
│ • Project Evaluation │
└─────────────────┬───────────────────────┘
│ Trigger
┌─────────────────▼───────────────────────┐
│ L0 Continuous Perception Layer│
│ (Every 5 min, perception) │
│ • Emotion Decay │
│ • Task Detection │
│ • Threshold Trigger │
└─────────────────────────────────────────┘
Responsibilities:
Output:
Consumption: ~800 tokens/run
Responsibilities:
Output:
Consumption: ~3000 tokens/run
Responsibilities:
Output:
Consumption: ~5000 tokens/run
Layer Communication: Via QUEUE.md markers
<!-- TRIGGER_L1: reason -->
<!-- TRIGGER_L2: reason -->
Trigger Conditions:
| Condition | Trigger |
|---|---|
| L0 emotion < threshold | Trigger L1 |
| L0 urgent task | Trigger L1 |
| L1 high-value project | Trigger L2 |
| L1 core file modification | Trigger L2 |
~/.openclaw/workspace/
├── SOUL.md # Core personality
├── PRINCIPLES.md # Decision principles
├── BRAIN.md # Brain Loop protocol
├── AUTONOMY.md # Autonomous rating system
├── EVOLUTION-FRAMEWORK.md # Evolution protocol
│
├── memory/
│ ├── inner-state.json # Emotion state
│ ├── core-confidence.json # Core file confidence
│ ├── autonomy-log.json # Autonomous execution record
│ └── dimensions-log.json # 5-dimension evolution
│
├── tasks/
│ └── QUEUE.md # Task queue + trigger markers
│
└── skills/
├── inner-life-quick/ # L0
├── inner-life-evolve-hourly/ # L1
└── inner-life-core/ # L2
# Copy templates to workspace
cp templates/SOUL.md ~/.openclaw/workspace/
cp templates/PRINCIPLES.md ~/.openclaw/workspace/
cp templates/BRAIN.md ~/.openclaw/workspace/
cp templates/AUTONOMY.md ~/.openclaw/workspace/
# Install Three-Layer Thinking Chain skills
openclaw skill install inner-life-quick
openclaw skill install inner-life-evolve-hourly
openclaw skill install inner-life-core
# L0: Every 5 minutes
openclaw cron add --cron "*/5 * * * *" --name "inner-life-quick" --session isolated
# L1: Every 1 hour
openclaw cron add --cron "0 * * * *" --name "inner-life-evolve-hourly" --session isolated
# L2: Every 4 hours
openclaw cron add --cron "0 */4 * * * *" --name "inner-life-brain" --session isolated
openclaw gateway restart
Edit SOUL.md:
Edit skills/inner-life-quick/SKILL.md:
Edit AUTONOMY.md:
| Layer | Frequency | Tokens/run | Daily consumption |
|---|---|---|---|
| L0 | 5 min | 800 | ~230k |
| L1 | 1 hour | 3000 | ~72k |
| L2 | 4 hours | 5000 | ~30k |
| Total | - | - | ~332k/day |
Recommendation: At least 400k tokens quota per day
✅ Suitable for:
❌ Not suitable for:
GitHub: https://github.com/openclaw/openclaw Discord: https://discord.com/invite/clawd
Designers: Yao + Saturday
Inspiration:
MIT
Version: 1.0.0 Created: 2026-03-13
Deploy KERNEL_INTUITION_Cp on any AI architecture with calibrated thresholds. Use when user wants to "activate Cp", "setup creative intuition", "deploy kernel on [AI]", "enable intuition mode", or mentions activating computational intuition on a specific model. Also trigger for "configure Cp for [architecture]" or "calibrate creative thresholds". This skill handles the full activation workflow including architecture profiling, threshold calibration, kernel generation, and monitoring setup.
CLI to manage emails via IMAP/SMTP. Use `himalaya` to list, read, write, reply, forward, search, and organize emails from the terminal. Supports multiple accounts and message composition with MML (MIME Meta Language).
OpenProse VM skill pack. Activate on any `prose` command, .prose files, or OpenProse mentions; orchestrates multi-agent workflows.
Delegate coding tasks to Codex, Claude Code, Cursor Agent, or Pi agents via background process. Use when: (1) building/creating new features or apps, (2) reviewing PRs (spawn in temp dir), (3) refactoring large codebases, (4) iterative coding that needs file exploration. NOT for: simple one-liner fixes (just edit), reading code (use read tool), thread-bound ACP harness requests in chat (for example spawn/run Codex or Claude Code in a Discord thread; use sessions_spawn with runtime:"acp"), or any work in ~/clawd workspace (never spawn agents here). Claude Code: use --print --permission-mode bypassPermissions (no PTY). Codex/Pi/OpenCode/Cursor Agent: pty:true required.
Host security hardening and risk-tolerance configuration for OpenClaw deployments. Use when a user asks for security audits, firewall/SSH/update hardening, risk posture, exposure review, OpenClaw cron scheduling for periodic checks, or version status checks on a machine running OpenClaw (laptop, workstation, Pi, VPS).
iMessage/SMS CLI for listing chats, history, and sending messages via Messages.app. Use when sending iMessages or SMS, reading chat history, or listing conversations. Trigger phrases: 'text', 'iMessage', 'send SMS', 'check messages'. NOT for: platforms with dedicated integrations (Discord, Telegram, Slack — use their skills), or when bluebubbles is configured (prefer that for iMessage).