← Back to Skills Marketplace
sipoon

Skill Compounding

by Sipoon · GitHub ↗ · v0.1.0 · MIT-0
cross-platform ⚠ suspicious
39
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install skill-compounding
Description
识别并提取可复用的解决方案为团队技能,实现经验积累与复利,避免重复摸索相同问题。
README (SKILL.md)

skill-compounding

借鉴来源:Multica (multica-ai/multica) 的 Reusable Skills 机制

每一个成功的解决方案,都应该成为团队未来可复用的技能。 技能随使用次数增加而积累,形成团队能力的复利。

核心理念:不要每次遇到相同问题都重新摸索。


触发条件

满足以下任一场景时激活:

  • 成功解决了一个非平凡问题
  • 发现了一个通用的解决方案(可以跨项目复用)
  • 用户明确要求"把这个流程固化下来"
  • 某个模式出现了 2 次以上

工作流程

Phase A:识别可复用的模式

问自己:

[compounding-check]
- 这个问题未来还会出现吗?
- 这个解决方案是否可以泛化?
- 这个流程是否有其他人需要?
- 沉淀为 Skill 的成本 vs 收益?

如果 >= 2 个"是",就值得沉淀。

Phase B:提取为 Skill

创建 Skill 目录结构:

skills/\x3Cskill-name>/
├── SKILL.md          # 核心文档(必须)
├── examples/         # 示例(可选)
├── templates/        # 模板(可选)
└── TOOLS.md         # 工具配置(可选)

SKILL.md 必须包含:

# \x3CSkill 名称>

## 何时使用
[一句话描述触发条件]

## 工作流程
[步骤 1]
[步骤 2]
[...]

## 输入/输出
- 输入:
- 输出:

## 限制与注意事项
[任何已知限制]

## 触发命令
[触发这个 Skill 的用户说法示例]

Phase C:注册到 Skill 索引

将新 Skill 添加到 skills/quick-reference.md(如果有),或者在 SOUL.md 的技能分层架构中声明。

Phase D:持续迭代

每次使用这个 Skill:

  • 如果发现问题 → 更新 SKILL.md
  • 如果有新场景 → 扩展 SKILL.md
  • 如果发现更优解 → 重写对应部分

Skill 生命周期

发现 → 提取 → 使用 → 迭代 → 传播
  ↑_____________________________|
        反馈驱动改进

质量标准

一个好的 Skill 应该满足:

标准 描述
可独立运行 不依赖其他 Skill 的上下文
触发条件明确 用户能说清楚什么时候该用它
验收标准清晰 完成后有可验证的结果
限制已说明 已知的不适用场景
有真实案例 至少验证过 1 次真实使用

与 Multica Skills 的对比

维度 Multica OpenClaw
存储位置 云端/团队共享 工作区 skills/
触发方式 Issue 路由 Skill 索引 + 手动
复用方式 团队共享 本地/可导出
积累可见性 团队面板 quick-reference.md

OpenClaw 的优势:更轻量,不需要服务器,直接文件系统管理。


触发命令

"把这个沉淀为 Skill"、"把这个流程记下来,以后复用"、"以后遇到 XXX 就用这个方案"

下一跳(Skill 链式调用)

skill-compounding 是技能沉淀技能,通常在其他技能链末端触发:

skill-compounding → 沉淀完成后回到原技能链

典型触发路径

  • \brainstorming Step 4 完成后 → 检查是否值得沉淀
  • conductor Implement 阶段交付后 → 检查是否有可复用成果
  • gent-teams 审查后发现通用模式 → 沉淀为 Skill
  • \refactoring 重构完成后 → 沉淀重构后的模式

不适用:纯一次性任务、商业敏感内容、个人偏好类内容。

Usage Guidance
Install only if you want the agent to help turn repeated work into reusable local skills. Before using it, require the agent to ask for confirmation before creating or registering any skill, and avoid capturing secrets, proprietary workflows, client data, or other sensitive project details.
Capability Assessment
Purpose & Capability
Creating skill folders, SKILL.md files, and index entries is coherent with the stated skill-compounding purpose.
Instruction Scope
Activation can occur after any non-trivial solved problem or repeated pattern, not only after an explicit user request, which makes the scope too broad for persistent workflow changes.
Install Mechanism
The package contains only markdown and JSON, with no executable scripts, dependencies, API keys, or network setup.
Credentials
The behavior is local and file-based, but it can preserve project procedures into reusable artifacts; the skill does warn against business-sensitive and personal-preference content.
Persistence & Privilege
The workflow directs the agent to create durable skills and update skill indexes or SOUL.md, which can affect future agent behavior without a clear confirmation or rollback step.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install skill-compounding
  3. After installation, invoke the skill by name or use /skill-compounding
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v0.1.0
- Initial release of the "skill-compounding" skill. - Adds a standardized workflow for identifying, extracting, and maintaining reusable skills within a team. - Includes clear triggers, a step-by-step process, and quality criteria for skill documentation. - Provides a directory structure and documentation template (SKILL.md) for new skills. - Outlines integration with skill indices and continuous iteration based on usage feedback. - Compares approach with Multica's team-based skill sharing and highlights OpenClaw’s lightweight, file-based management.
Metadata
Slug skill-compounding
Version 0.1.0
License MIT-0
All-time Installs 0
Active Installs 0
Total Versions 1
Frequently Asked Questions

What is Skill Compounding?

识别并提取可复用的解决方案为团队技能,实现经验积累与复利,避免重复摸索相同问题。 It is an AI Agent Skill for Claude Code / OpenClaw, with 39 downloads so far.

How do I install Skill Compounding?

Run "/install skill-compounding" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is Skill Compounding free?

Yes, Skill Compounding is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does Skill Compounding support?

Skill Compounding is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created Skill Compounding?

It is built and maintained by Sipoon (@sipoon); the current version is v0.1.0.

💬 Comments