Claude Code持续学习技能：自动提取会话模式，提升AI编程效率

continuous-learning by affaan-m/everything-claude-code

1,600 周安装量

102,100 GitHub Stars

GitHub

安装命令

npx skills add https://github.com/affaan-m/everything-claude-code --skill continuous-learning

AI/机器学习开发生产力

🇨🇳中文介绍

持续学习技能

自动评估 Claude Code 会话的结尾，提取可重复使用的模式，并将其保存为已学习的技能。

何时激活

设置从 Claude Code 会话自动提取模式时
配置用于会话评估的 Stop 钩子时
在 ~/.claude/skills/learned/ 中查看或管理已学习的技能时
调整提取阈值或模式类别时
比较 v1（当前版本）与 v2（基于本能）的方法时

工作原理

此技能作为 Stop 钩子 在每个会话结束时运行：

会话评估：检查会话是否包含足够多的消息（默认：10 条以上）
模式检测：从会话中识别可提取的模式
技能提取：将有用的模式保存到 ~/.claude/skills/learned/

配置

编辑 config.json 以进行自定义：

{
  "min_session_length": 10,
  "extraction_threshold": "medium",
  "auto_approve": false,
  "learned_skills_path": "~/.claude/skills/learned/",
  "patterns_to_detect": [
    "error_resolution",
    "user_corrections",
    "workarounds",
    "debugging_techniques",
    "project_specific"
  ],
  "ignore_patterns": [
    "simple_typos",
    "one_time_fixes",
    "external_api_issues"
  ]
}

广告位招租

在这里展示您的产品或服务

触达数万 AI 开发者，精准高效

联系我们

模式	描述
`error_resolution`	特定错误的解决方式
`user_corrections`	来自用户纠正的模式
`workarounds`	针对框架/库特殊情况的解决方案
`debugging_techniques`	有效的调试方法
`project_specific`	项目特定的约定

为什么使用 Stop 钩子？

轻量级：仅在会话结束时运行一次
非阻塞：不会给每条消息增加延迟
完整上下文：可以访问完整的会话记录

长篇指南 - 关于持续学习的章节
/learn 命令 - 在会话中手动提取模式

对比说明（研究时间：2025年1月）

与 Homunculus 对比

Homunculus v2 采用了更复杂的方法：

特性	我们的方法	Homunculus v2
观察方式	Stop 钩子（会话结束时）	PreToolUse/PostToolUse 钩子（100% 可靠）
分析方式	主上下文	后台代理（Haiku）
粒度	完整技能	原子化的“本能”
置信度	无	0.3-0.9 加权
演化路径	直接成为技能	本能 → 聚类 → 技能/命令/代理
共享	无	导出/导入本能

来自 homunculus 的关键见解：

"v1 依赖技能进行观察。技能是概率性的——它们触发的概率约为 50-80%。v2 使用钩子进行观察（100% 可靠），并使用本能作为学习行为的原子单位。"

潜在的 v2 增强功能

基于本能的学习 - 更小、原子化的行为，带有置信度评分
后台观察器 - Haiku 代理并行分析
置信度衰减 - 如果被反驳，本能会降低置信度
领域标记 - 代码风格、测试、git、调试等
演化路径 - 将相关本能聚类成技能/命令

完整规范请参见：docs/continuous-learning-v2-spec.md。

🇺🇸English

Continuous Learning Skill

Automatically evaluates Claude Code sessions on end to extract reusable patterns that can be saved as learned skills.

When to Activate

Setting up automatic pattern extraction from Claude Code sessions
Configuring the Stop hook for session evaluation
Reviewing or curating learned skills in ~/.claude/skills/learned/
Adjusting extraction thresholds or pattern categories
Comparing v1 (this) vs v2 (instinct-based) approaches

How It Works

This skill runs as a Stop hook at the end of each session:

Session Evaluation : Checks if session has enough messages (default: 10+)
Pattern Detection : Identifies extractable patterns from the session
Skill Extraction : Saves useful patterns to ~/.claude/skills/learned/

Configuration

Edit config.json to customize:

{
  "min_session_length": 10,
  "extraction_threshold": "medium",
  "auto_approve": false,
  "learned_skills_path": "~/.claude/skills/learned/",
  "patterns_to_detect": [
    "error_resolution",
    "user_corrections",
    "workarounds",
    "debugging_techniques",
    "project_specific"
  ],
  "ignore_patterns": [
    "simple_typos",
    "one_time_fixes",
    "external_api_issues"
  ]
}

Pattern Types

Pattern	Description
`error_resolution`	How specific errors were resolved
`user_corrections`	Patterns from user corrections
`workarounds`	Solutions to framework/library quirks
`debugging_techniques`	Effective debugging approaches
`project_specific`	Project-specific conventions

Hook Setup

Add to your ~/.claude/settings.json:

{
  "hooks": {
    "Stop": [{
      "matcher": "*",
      "hooks": [{
        "type": "command",
        "command": "~/.claude/skills/continuous-learning/evaluate-session.sh"
      }]
    }]
  }
}

Why Stop Hook?

Lightweight : Runs once at session end
Non-blocking : Doesn't add latency to every message
Complete context : Has access to full session transcript

The Longform Guide - Section on continuous learning
/learn command - Manual pattern extraction mid-session

Comparison Notes (Research: Jan 2025)

vs Homunculus

Homunculus v2 takes a more sophisticated approach:

Feature	Our Approach	Homunculus v2
Observation	Stop hook (end of session)	PreToolUse/PostToolUse hooks (100% reliable)
Analysis	Main context	Background agent (Haiku)
Granularity	Full skills	Atomic "instincts"
Confidence	None	0.3-0.9 weighted
Evolution	Direct to skill	Instincts → cluster → skill/command/agent
Sharing	None	Export/import instincts

Key insight from homunculus:

"v1 relied on skills to observe. Skills are probabilistic—they fire ~50-80% of the time. v2 uses hooks for observation (100% reliable) and instincts as the atomic unit of learned behavior."

Potential v2 Enhancements

Instinct-based learning - Smaller, atomic behaviors with confidence scoring
Background observer - Haiku agent analyzing in parallel
Confidence decay - Instincts lose confidence if contradicted
Domain tagging - code-style, testing, git, debugging, etc.
Evolution path - Cluster related instincts into skills/commands

See: docs/continuous-learning-v2-spec.md for full spec.

Weekly Installs

1.0K

Repository

affaan-m/everyt…ude-code

GitHub Stars

70.6K

First Seen

Jan 22, 2026

Security Audits

Gen Agent Trust HubPass SocketPass SnykPass

Installed on

opencode838

codex812

gemini-cli796

claude-code785

github-copilot712

cursor701