← 返回 Skills 市场
Rag Evaluator
作者
bytesagain1
· GitHub ↗
· v2.0.0
· MIT-0
191
总下载
0
收藏
1
当前安装
1
版本数
在 OpenClaw 中安装
/install rag-evaluator
功能描述
Python SDK for Agent AI Observability, Monitoring and Evaluation Framework. Includes features like a ragaai catalyst, python, agentic-ai.
安全使用建议
This skill appears coherent and local-only: it stores logs and exports under ~/.local/share/rag-evaluator and does not contact external endpoints or request credentials. Before installing or running it, review the included scripts (scripts/script.sh). Be aware that commands like `prompt`, `configure`, or `evaluate` log whatever you pass — avoid entering secrets or API keys into those entries. Exports and logs are stored on disk (they may grow over time), so consider disk space and file permissions. If you want extra safety, run the script in an isolated environment (container or dedicated user account) or modify DATA_DIR to point to a controlled location. Finally, note the JSON export is implemented naively and may produce invalid JSON for values containing quotes or newlines; treat exported files accordingly.
功能分析
Type: OpenClaw Skill
Name: rag-evaluator
Version: 2.0.0
The skill is a local logging utility for RAG (Retrieval-Augmented Generation) evaluation metrics, implemented as a Bash script. It records user-provided strings into categorized log files within `~/.local/share/rag-evaluator/` and provides basic search and export functionality. While the documentation in SKILL.md incorrectly refers to it as a 'Python SDK', the actual script (scripts/script.sh) is transparent, lacks network access, and contains no indicators of malicious intent, data exfiltration, or unauthorized execution.
能力评估
Purpose & Capability
Name/description (Rag Evaluator) match the behavior: a small CLI that logs evaluation/benchmark/prompt data and provides exports, search, stats, and status. There are no unrelated env vars, binaries, or services requested.
Instruction Scope
Runtime instructions and the included script only read/write files in ~/.local/share/rag-evaluator and run standard Unix utilities. However, the tool logs arbitrary user-provided inputs (prompts, configs, costs, etc.) so users could accidentally store sensitive secrets or API keys. The JSON export implementation is brittle (naive string formatting, no escaping, and uses a literal "\n]" which may produce invalid JSON), so exported data may be malformed.
Install Mechanism
There is no install spec (instruction-only skill) which is low risk. A shell script is included; it is plain Bash and not obfuscated. Because there's no automatic installer, the script won't be placed on PATH by the platform unless the agent/user does so; inspect the script before running or install into an isolated environment if desired.
Credentials
The skill declares no required environment variables or credentials and the script only uses HOME and standard utilities. Requested accesses are proportional to a local logging CLI.
Persistence & Privilege
The script persists logs and exports under the user's home (~/.local/share/rag-evaluator) which is expected for this tool. It does not request elevated privileges, does not modify other skills, and always:false. Consider the privacy implications of storing inputs on disk.
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 rag-evaluator - 安装完成后，直接呼叫该 Skill 的名称或使用
/rag-evaluator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
v2.0.0
publish v2.0.0
元数据
常见问题
Rag Evaluator 是什么？
Python SDK for Agent AI Observability, Monitoring and Evaluation Framework. Includes features like a ragaai catalyst, python, agentic-ai. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 191 次。
如何安装 Rag Evaluator？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install rag-evaluator」即可一键安装，无需额外配置。
Rag Evaluator 是免费的吗？
是的，Rag Evaluator 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Rag Evaluator 支持哪些平台？
Rag 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 Rag Evaluator？
由 bytesagain1（@bytesagain1）开发并维护，当前版本 v2.0.0。
推荐 Skills