Skill Reviewer
v1.0.0审查和评估 OpenClaw Skill 的质量,提供改进建议
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
Name/description claim to review other skills; the runtime instructions only require reading a target SKILL.md and producing a checklist-based review. No unrelated credentials, binaries, or installs are requested.
Instruction Scope
SKILL.md explicitly tells the agent to read the target skill's SKILL.md (example cat ~/.openclaw/skills/... or /opt/homebrew/...). That is expected for a reviewer, but it means the agent will access local skill files — which could contain sensitive notes if authors put secrets there. The instructions do not ask for other system files, env vars, or external endpoints.
Install Mechanism
No install spec and no code files (instruction-only). This minimizes on-disk execution risk.
Credentials
No environment variables, credentials, or config paths are declared or required. The requested access (reading SKILL.md) aligns with the stated purpose.
Persistence & Privilege
always:false and normal invocation model. The skill does not request permanent presence or system-wide config changes.
Assessment
This skill appears coherent: it reviews other skills by reading their SKILL.md files and returns checklist-based feedback. Before installing, consider: (1) the agent will be able to read local skill files — ensure you don't store secrets in SKILL.md; (2) prefer keeping the skill user-invocable (not always-enabled) so it runs only when you ask; (3) review sample outputs and test it on non-sensitive skills first. If you want stronger guarantees, inspect the skill text and run it with least privilege or in a sandboxed environment.Like a lobster shell, security has layers — review code before you run it.
Runtime requirements
🔍 Clawdis
analysislatestreview
Skill 质量审查器
审查一个 Skill 的质量,判断其是否是个好 Skill。
When to Use
✅ USE this skill when:
- 用户想审查某个 Skill 的质量
- 评估一个 Skill 是否需要改进
- 检查 Skill 是否符合最佳实践
- 给 Skill 打分或提出改进建议
❌ DON'T use this skill when:
- 用户只是询问某个 Skill 的使用方法
- 需要直接修改某个 Skill(只负责审查,不负责修改)
- 与 Skill 开发无关的一般性问题
什么是好的 Skill
1. 描述精准
- ✅ 清晰说明能做什么
- ❌ 避免太宽泛(如"查东西"、"GitHub 相关")
2. 场景边界清晰
- 有 When to Use(什么时候用)
- 有 When NOT to Use(什么时候不用)
3. 命令具体可执行
- 有可直接运行的命令示例
- ❌ 不能只有概念,没有具体实现
4. 专注一件事
- 一次只做一件事,窄而深
- ❌ 避免瑞士军刀式的大杂烩
审查清单
给目标 Skill 逐项检查:
| 检查项 | 权重 | 说明 |
|---|---|---|
| Header 完整 | ⭐⭐⭐ | 包含 name, description, metadata |
| description 精准 | ⭐⭐⭐⭐⭐ | 不模糊,不宽泛 |
| 有明确 Use Cases | ⭐⭐⭐⭐ | 有 When to Use 说明 |
| 有具体命令示例 | ⭐⭐⭐⭐⭐ | 可直接复制运行 |
| 有注意事项 | ⭐⭐⭐ | 包含 limitations, requirements |
| 专注一件事 | ⭐⭐⭐⭐ | 不做"大杂烩" |
评分标准
| 等级 | 分数 | 评价 |
|---|---|---|
| 优秀 | 90-100 | 可直接使用,无需修改 |
| 良好 | 70-89 | 不错,稍作优化更佳 |
| 一般 | 50-69 | 需要较大改进 |
| 较差 | <50 | 建议重新设计 |
注意事项
- ⚠️ 审查结果仅供参考,最终判断需人工决定
- ⚠️ 只负责审查,不自动修改目标 Skill
- ⚠️ 需要先读取目标 SKILL.md 才能审查(无法审查不存在的文件)
- 💡 建议结合具体使用场景灵活应用
实际示例
审查步骤
-
读取目标 Skill 的 SKILL.md
# 假设要审查 weather skill cat ~/.openclaw/skills/weather/SKILL.md # 或 cat /opt/homebrew/lib/node_modules/openclaw/skills/weather/SKILL.md -
逐项对照审查清单检查
-
输出审查结果(见下方模板)
示例输出
## 审查结果:weather skill
### 得分:95/100 (优秀)
### 优点
- Header 完整,包含 emoji
- description 精准,不模糊
- 有完整的 When to Use / When NOT to Use
- 命令示例丰富且可直接运行
- 包含注意事项(rate limit)
### 问题
- 无明显问题
### 改进建议
- 可考虑添加更多城市的格式示例
输出格式
审查完成后,输出:
## 审查结果:[Skill 名称]
### 得分:X/100 (等级)
### 优点
- ...
### 问题
- ...
### 改进建议
1. ...
2. ...
使用方法
当需要审查某个 Skill 的质量时:
- 读取该 Skill 的 SKILL.md 文件
- 根据上述清单逐项检查
- 输出审查结果和改进建议
Comments
Loading comments...
