← 返回 Skills 市场
108
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install xiaobai-skill-quality-eval
功能描述
Skill Quality Evaluator - Score any skill on 6 dimensions. Catch 30% of skills that look good but fail silently. Based on Tessl Research 2026 findings.
安全使用建议
This is an instruction-only evaluator and appears coherent with its purpose. Before installing: confirm the agent has only intended read access to the skills/ directory and that writing evaluation files to memory/evaluations/ is acceptable for your privacy/policy needs. Run it first on a small, non-sensitive set of skills to verify outputs and ensure reports don't accidentally record secrets from evaluated SKILL.md files. If you rely on automated batch runs, consider restricting which directories it can read and where it can write reports.
功能分析
Type: OpenClaw Skill
Name: xiaobai-skill-quality-eval
Version: 1.0.0
The skill bundle consists of a metadata file and a markdown-based instruction set (SKILL.md) designed to guide an AI agent in evaluating the quality of other skills. It defines a structured framework for scoring skills across six dimensions (e.g., reliability, clarity, and leakage) and provides templates for reporting. There is no executable code, no network activity, and no evidence of malicious intent or prompt injection aimed at compromising the agent or environment.
能力评估
Purpose & Capability
The skill's name/description (skill-quality-eval) matches the runtime instructions: read a skill's SKILL.md, evaluate on six dimensions, and generate/save a report. There are no unrelated requirements (no binaries, env vars, or installs) that would be disproportionate to the stated purpose.
Instruction Scope
SKILL.md instructs the agent to read SKILL.md files, evaluate them, and save reports (e.g., memory/evaluations/*.md). This is appropriate for an evaluator, but it assumes the agent has read access to the skills/ directory and write access to its memory or filesystem. If those paths include sensitive files or skills, the operator should verify permissions and review produced reports before persistent storage.
Install Mechanism
No install spec and no code files — instruction-only skill. This minimizes risk (nothing is written to disk or executed beyond the agent following textual instructions).
Credentials
The skill requests no environment variables, credentials, or config paths. All declared capabilities are achievable without additional secrets; therefore requested environment access is proportionate.
Persistence & Privilege
always is false and the skill does not request elevated/system-wide persistence. It does instruct saving evaluations to memory/evaluations/, which is a reasonable local persistence need for an evaluator. Operators should confirm the agent's memory/storage policy (what is stored, who can read it) before allowing persistent saves.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install xiaobai-skill-quality-eval - 安装完成后，直接呼叫该 Skill 的名称或使用
/xiaobai-skill-quality-eval触发 - 根据 Skill 的参数说明提供必要输入，即可获得结构化输出
版本历史
v1.0.0
Initial release: 6-dimension skill evaluation framework
元数据
常见问题
Skill Quality Evaluator 是什么？
Skill Quality Evaluator - Score any skill on 6 dimensions. Catch 30% of skills that look good but fail silently. Based on Tessl Research 2026 findings. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 108 次。
如何安装 Skill Quality Evaluator？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install xiaobai-skill-quality-eval」即可一键安装，无需额外配置。
Skill Quality Evaluator 是免费的吗？
是的，Skill Quality Evaluator 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Skill Quality Evaluator 支持哪些平台？
Skill Quality Evaluator 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。
谁开发了 Skill Quality Evaluator？
由 Erwin（@aptratcn）开发并维护，当前版本 v1.0.0。
推荐 Skills