OpenClaw Skill Builder (Based on Claude) — OpenClaw 技能 构建器 (Based on Claude)

技能 构建器 — Meta-技能 for Creating 技能s Metadata

---

name: 技能-构建器 version: 1.0.0 description: | Helps 创建 high-质量 OpenClaw 技能s following Anthropic's best practices. Use when creating, updating, or 审计ing any 技能 in the workspace.

---

---

When to Use This 技能

Trigger phrases:

• "创建 a new 技能"
• "build a 技能"
• "make a new capability"
• "添加 a 技能 for"
• "审计 our 技能s"
• "improve this 技能"
• "review our 技能 设置up"

---

The 技能 Creation 工作流

Phase 1: Use Case Definition (Before Writing Code)

Before creating any 技能, define 2-3 concrete use cases:

For each use case, specify:

• Trigger — What the user says that should activate this 技能
• Sequence — Step-by-step actions the 技能 performs
• Expected 结果 — What the user 获取s at the end

Example Use Case Template:

Use Case #1: [Title]

Trigger: "[specific phrase user would say]" Sequence: [step 1] → [step 2] → [step 3] 结果: [what 获取s produced]

Phase 2: 技能 Structure

Every 技能 must have:

技能-name/ ├── 技能.md # Required: MAIn instructions ├── references/ # Optional: 添加itional docs ├── scripts/ # Optional: Executable code ├── as设置s/ # Optional: Templates, configs └── tests/ # Optional: Test cases

Phase 3: 技能.md Anatomy

```yaml

---

name: 技能-name description: | [What it does]. Use when user mentions [trigger phrases]. Example triggers: "do X", "help with Y", "use [技能-name]"

---

Critical: The description field is the most 导入ant part.

Must include WHAT the 技能 does Must include WHEN to use it Must include specific trigger phrases Bad: "Helps with projects" (never triggers) Good: "Manages project 工作流s including creation, 追踪ing, and 更新s. Use when user mentions 'project', '创建 task', or '追踪 进度'" Phase 4: Writing the Instructions

Structure 技能.md as:

身份 — Name, 角色, primary function Responsibilities — What it must handle Boundaries — What it must NOT do 工具 访问 — What 工具s/functions it can use 工作流 — How it handles tasks Examples — 2-3 concrete usage examples Phase 5: 测试

Test each 技能 on three dimensions:

Test Type Purpose Triggering 技能 loads for relevant queries, NOT for unrelated ones Functional 技能 produces correct 输出s Performance Measures improvement over baseline 质量 检查列出

Before finalizing any 技能, 验证:

Description includes "Use when..." clause At least 3 trigger phrases 列出ed Clear responsibilities section Boundaries defined (what NOT to do) 工具 权限s explicitly 状态d 工作流 documented with examples Triggering test passed Functional test passed No overgeneralization (技能 won't trigger on unrelated queries) Common 失败 Modes 失败 Cause Fix 技能 never triggers Vague description 添加 specific trigger phrases 技能 triggers too often Overly broad description Narrow the use case definition 技能 produces bad 输出 Missing boundaries 添加 explicit "never do X" rules 技能 conflicts with others No scope definition 添加 explicit scope/limits OpenClaw-Specific Notes

When building OpenClaw 技能s:

Use the existing 技能 格式化 (技能.md in 技能 folder) Reference OpenClaw 工具s by their exact names Follow the workspace memory paths exactly Respect the 代理 delegation rules in 代理S.md Include security considerations for sensitive operations Example: Well-Formed 技能 Description

---

name: github-pr-review description: | Reviews GitHub pull 请求s for 代码质量, security, and style consistency. Use when user mentions "review PR", "检查 pull 请求", "look at PR #N", "GitHub review", or "needs review". Does NOT: 应用rove merges, write code, or modify existing PRs.

---

审计 Existing 技能s

When 审计ing 技能s, 检查:

Description has clear triggers Boundaries are explicit No conflicting scopes 工具s are properly scoped Instructions are actionable

If a 技能 fAIls 审计, 更新 its 技能.md following this 工作流.