Skip to content
Skills Library
AI & agents · Automation

continuous-learning

/continuous-learning

自动从 Claude Code 会话中提取可重用模式,并将其保存为 learned skills 供将来使用。

Licence
MIT
Last repository change
21.06.2026

Assessment and security

96/ 100Class: good
Format25 / 25
Frontmatter with name and description, valid name, description up to 1,024 characters, name matching the folder.
Discoverability21 / 25
Description explains when to use the skill, has a reasonable length and uses third person.
Content35 / 35
At least 150 words and at most 500 lines, headings, step lists, examples or code.
Hygiene15 / 15
No placeholders such as TODO, no Windows paths, with a licence.
Riskno signals detected in the scan
Files in this package
3
Scripts
1
HTML comments
0
URLs in SKILL.md
1
Deprecation

No signals detected. Repository changed on 21.06.2026, without references to retired models or APIs.

Licence

MIT, open: use and modify while retaining author attribution.

What to check before installing
  • 1 scripts in this package. Review them before installing, as Claude may execute them.

Scripts: evaluate-session.sh

What it does and when to use it

自动从 Claude Code 会话中提取可重用模式,并将其保存为 learned skills 供将来使用。

Original author description

自动从 Claude Code 会话中提取可重用模式,并将其保存为 learned skills 供将来使用。

How it works

  1. 工作原理
  2. 配置
  3. 模式类型
  4. Hook 设置
  5. 为什么使用 Stop Hook?
  6. 相关
  7. 比较说明 (Research: Jan 2025)
Original SKILL.md headings · 107 lines, 284 words
SKILL.md preview
---
name: continuous-learning
description: 自动从 Claude Code 会话中提取可重用模式,并将其保存为 learned skills 供将来使用。
origin: ECC
---

# 持续学习技能 (Continuous Learning Skill)

自动在结束时评估 Claude Code 会话,以提取可重用模式并将其保存为 learned skills。

## 工作原理

此技能作为 **Stop hook** 在每个会话结束时运行:

1. **会话评估**: 检查会话是否有足够的消息 (默认: 10+)
2. **模式检测**: 识别会话中可提取的模式
3. **技能提取**: 将有用的模式保存到 `~/.claude/skills/learned/`

## 配置

编辑 `config.json` 进行自定义:

```json
{
  "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"
  ]
}
```

## 模式类型

| Pattern | Description |
|---------|-------------|
| `error_resolution` | 特定错误是如何解决的 |
| `user_corrections` | 来自用户纠正的模式 |
| `workarounds` | 框架/库怪癖的解决方案 |
| `debugging_techniques` | 有效的调试方法 |
| `project_specific` | 项目特定约定 |

## Hook 设置

添加到你的 `~/.claude/settings.json`:

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

## 为什么使用 Stop Hook?

- **轻量级**: 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

## 相关

- [The Longform Guide](https://x.com/affaanmustafa/status/2014040193557471352) - 持续学习章节
- `/learn` 命令 - 会话中手动提取模式

---

## 比较说明 (Research: Jan 2025)

### vs Homunculus (github.com/humanplane/homunculus)

Homunculus v2 采取了更复杂的方法:

| Feature | Our Approach | Homunculus v2 |
|---------|--------------|---------------|
| 观察 (Observation) | Stop hook (会话结束) | PreToolUse/PostToolUse hooks (100% 可靠) |
| 分析 (Analysis) | 主上下文 | 后台 agent (Haiku) |
| 粒度 (Granularity) | 完整技能 | 原子 "instincts" (本能) |
| 置信度 (Confidence) | 无 | 0.3-0.9 加权 |
| 演进 (Evolution) | 直接从 skill | Instincts → cluster → skill/command/agent |
| 共享 (Sharing) | 无 | 导出/导入 instincts |

**来自 homunculus 的关键见解:**
> "v1 依靠技能进行观察。技能是概率性的——它们大约 50-80% 的时间触发。v2 使用 hooks 进行观察(100% 可靠)并将 instincts 作为学习行为的原子单位。"

### 潜在 v2 增强功能

1. **基于 Instinct 的学习** - 具有置信度评分的更小、原子行为
2. **后台观察者** - 并行分析的 Haiku agent
3. **置信度衰减** - 如果 instincts 被反驳,置信度会降低
4. **领域标记** - code-style, testing, git, debugging, etc.
5. **演进路径** - 将相关的 instincts 聚类为 skills/commands

参见: `/Users/affoon/Documents/tasks/12-continuous-learning-v2.md` 获取完整规范。
Files in this package (3)
  • SKILL.md 3 KB
  • config.json 1 KB
  • evaluate-session.sh 2 KB

Versions

RepositoryScoreLicence
This page · aaione/everything-claude-code-zh96MIT
ericrisco/rsc-harness100MIT
tadokoro-ryusuke/cc-plugins100MIT
mhylle/claude-skills-collection100MIT
ckorhonen/claude-skills100MIT
affaan-m/ECC96MIT
affaan-m/ECC96MIT
mturac/everything-openai-codex96MIT
mturac/everything-openai-codex96MIT

Database snapshot: 13.11.2026 · package version 80b5fd3bda45 (SHA-256 digest of the entire package)

Downloads in the last 30 days: 0 · ranking

Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.

Similar skills

  • 100continuous-learning-v2
    Instinct-based learning system with confidence scoring, import/export, and evolution into skills. The system learns patterns from sessions and promotes high-confidence instincts into durable skills. Use when extracting patterns from completed work.
  • 100session-learning
    This skill should be used when the user asks about "session learning", "learn from session", "improve future sessions", "session reflection", "session metrics", "session analysis", "/session-reflect", "req learning", "learning updates", "rollback learning", or wants to…
  • 100continuous-learning-v2
    Instinct-based learning system that observes sessions via hooks, creates atomic instincts with confidence scoring, and evolves them into skills/commands/agents. v2.1 adds project-scoped instincts to prevent cross-project contamination. Use when capturing lessons from a session,…
  • 100skill-factory
    Analyze session work and automatically convert reusable patterns into Claude Code skills. Use when: "세션을 스킬로", "스킬 만들어", "이거 스킬로", "skill factory", "이 작업 자동화해", "스킬 추출", "make this a skill", "extract skill", "convert to skill", "스킬 팩토리", "자동 스킬 생성". Differs from skill-creator…
  • 100self-improving-agent
    Curate Claude Code's auto-memory into durable project knowledge. Analyze MEMORY.md for patterns, promote proven learnings to CLAUDE.md and .claude/rules/, extract recurring solutions into reusable skills. Use when: (1) reviewing what Claude has learned about your project, (2)…
  • 95writing-claude-skills
    Use when user asks to create, write, edit, or test a skill. Also use when documenting reusable techniques, patterns, or workflows for future Claude instances.

Feedback and reports

Your message goes to Bartosz Sokoliński. An email address is needed only for a reply. See the privacy policy.