Self-Evolving Skill – OpenClaw Skill

---

name: Self-Evolving Skill description: Meta-cognitive self-learning system - Automated skill evolution based on predictive coding and value-driven mechanisms. homepage: https://github.com/whtoo/self-evolving-bot

---

Self-Evolving Skill

元认知自学习系统 - 基于预测编码和价值驱动的Skill自动演化。

功能

• ResidualPyramid金字塔分解，量化认知缺口 -: 残差 自适应反思触发: 基于残差能量自动判断何时需要学习
• 经验回放: 缓存已学模式，降低重复触发
• 价值门控: 只有提升长期价值才接受变异
• 持久化: 经验自动保存/加载

安装

# 技能已安装到 ~/.openclaw/skills/self-evolving-skill
# 或使用ClawHub
clawhub install self-evolving-skill

架构

self-evolving-skill/
├── core/                      # Python核心
│   ├── residual_pyramid.py     # 残差金字塔（SVD分解）
│   ├── reflection_trigger.py  # 自适应触发器
│   ├── experience_replay.py   # 经验回放缓存
│   ├── skill_engine.py        # 核心引擎+ValueGate
│   ├── storage.py             # 持久化
│   └── mcp_server.py          # MCP服务器
├── src/                       # TypeScript SDK
│   ├── index.ts               # 主入口
│   ├── cli.ts                 # CLI
│   └── mcp-tools.ts           # 工具定义
├── skills/                    # OpenClaw Skill
│   └── self-evolving-skill/    # 技能封装
├── MCP_CONFIG.md              # MCP配置
└── README.md                   # 文档

MCP工具

工具	描述	参数
skill_create	创建Skill	name, description
skill_execute	执行并学习	skill_id, context, success, value
skill_analyze	分析嵌入	embedding
skill_list	列出Skills	-
skill_stats	系统统计	-
skill_save	持久化保存	skill_id
skill_load	加载	skill_id

使用方式

CLI

# 列出所有Skill
openclaw skill self-evolving-skill list

# 创建Skill
openclaw skill self-evolving-skill create --name "MySkill"

# 执行
openclaw skill self-evolving-skill execute <id> --success

# 分析
openclaw skill self-evolving-skill analyze --embedding '[0.1,0.2,...]'

# 统计
openclaw skill self-evolving-skill stats

MCP服务器

# 启动MCP服务器
cd ~/.openclaw/skills/self-evolving-skill
./run_mcp.sh

# 或使用适配器
python3 mcporter_adapter.py skill_list '{}'

编程

import { SelfEvolvingSkillEngine } from 'self-evolving-skill';

const engine = new SelfEvolvingSkillEngine();
await engine.init();

const { skillId } = await engine.createSkill({ name: 'Analyzer' });
const stats = await engine.stats();

核心算法

1. 残差金字塔分解

pyramid = ResidualPyramid(max_layers=5, use_pca=True)
decomposition = pyramid.decompose(embedding)

# 输出:
# - residual_ratio: 残差能量比率
# - suggested_abstraction: POLICY / SUB_SKILL / PREDICATE
# - novelty_score: 综合新颖性

2. 三层跃迁规则

覆盖率	抽象层级	操作
>80%	POLICY	调整策略权重
40-80%	SUB_SKILL	生成子Skill
<40%	PREDICATE	归纳新谓词

3. 自适应阈值

trigger = ReflectionTrigger(
  min_energy_ratio=0.10,     # 初始阈值
  value_gain_threshold=0.20, # 触发阈值
  target_trigger_rate=0.15   # 目标15%触发率
)

文件位置

路径	说明
~/.openclaw/skills/self-evolving-skill	技能根目录
~/.openclaw/mcp_servers/self-evolving-skill.json	MCP服务器配置
~/.openclaw/workspace/self-evolving-skill/storage	数据存储

相关文档

• README.md - 完整文档
• MCP_CONFIG.md - MCP配置说明
• MEMORY.md - 研究笔记

Self-Evolving Skill - OpenClaw集成

项目结构

self-evolving-skill/
├── core/                    # Python核心模块
│   ├── residual_pyramid.py   # 残差金字塔分解
│   ├── reflection_trigger.py  # 自适应触发器
│   ├── experience_replay.py  # 经验回放
│   ├── skill_engine.py       # 核心引擎
│   ├── storage.py           # 持久化
│   └── mcp_server.py        # MCP服务器
├── src/                     # TypeScript封装
│   ├── index.ts            # 主入口
│   ├── cli.ts              # CLI
│   └── mcp-tools.ts        # MCP工具定义
├── skills/                 # 供OpenClaw调用
│   └── self-evolving-skill/  # OpenClaw Skill
├── SKILL.md                # 技能文档
├── package.json
└── README.md

安装到OpenClaw

# 方式1: 链接到OpenClaw skills目录
cd skills/self-evolving-skill
npm install
npm run build

# 链接
ln -s $(pwd)/skills/self-evolving-skill ~/.openclaw/skills/self-evolving-skill

# 方式2: 通过ClawHub
clawhub install self-evolving-skill

OpenClaw中调用

// 直接调用MCP工具
const result = await useTool('skill_create', {
  name: 'ProblemSolver'
});

const analysis = await useTool('skill_analyze', {
  embedding: [0.1, 0.2, 0.3, ...]
});

MCP工具列表

工具	描述	参数
skill_create	创建Skill	name, description
skill_execute	执行并学习	skill_id, context, success
skill_analyze	分析嵌入	embedding
skill_list	列出Skills	-
skill_stats	系统统计	-
skill_save	持久化保存	skill_id
skill_load	加载	skill_id

示例

// 1. 创建Skill
const skill = await useTool('skill_create', {
  name: 'TextAnalyzer',
  description: '文本分析自学习Skill'
});

// 2. 执行并观察学习
const result = await useTool('skill_execute', {
  skill_id: skill.skill_id,
  context: { task: 'sentiment' },
  success: true,
  value: 1.0
});

console.log('反思触发:', result.reflection_triggered);

// 3. 分析新输入
const analysis = await useTool('skill_analyze', {
  embedding: generateEmbedding(text)
});

配置

在OpenClaw配置文件中：

skills:
  self-evolving-skill:
    max_layers: 5
    energy_threshold: 0.1
    similarity_threshold: 0.85
    target_trigger_rate: 0.15
    storage_dir: ~/.openclaw/self-evolving