← 返回 Skills 市场
Evaluation Suite
作者
yuyonghao-123
· GitHub ↗
· v0.1.2
· MIT-0
233
总下载
0
收藏
0
当前安装
7
版本数
在 OpenClaw 中安装
/install yuyonghao-evaluation-suite
功能描述
Provides API for evaluating RAG quality, logical reasoning, and detecting hallucinations in AI-generated content with batch support.
安全使用建议
This package appears coherent and implements simple heuristic evaluators; there is no network exfiltration or secret access. Before installing/using: (1) review the API parameter names — the SKILL.md examples and the module function signatures differ, so adapt your integration to the code (e.g., use { text, sources, query } for hallucination detector and { query, answer, contexts } for RAG). (2) Run npm test and the quick-test in an isolated/sandboxed environment first; package.json lacks dependencies but some scripts reference missing files (benchmarks) and Node ESM/CommonJS mismatches may cause runtime errors (package.json lacks "type":"module"). (3) Because the source and homepage are unknown, audit the code for logic correctness on representative inputs if you plan to use results for evaluation decisions. If you need stronger guarantees, ask the publisher for provenance (source repo, release tags) or a signed/reproducible release.
能力评估
Purpose & Capability
Name/description (RAG, reasoning, hallucination detection) match the shipped JavaScript modules. The bundled files implement the stated functionality and do not require unrelated services or credentials. Minor provenance issues: source/homepage unknown and package author is an alias, but this is metadata not a capability mismatch.
Instruction Scope
SKILL.md describes how to use the API and to run npm install/npm test. However there are several small inconsistencies between docs and code: SKILL.md examples use different parameter names (e.g., generatedText/context/retrievedDocs) than the code expects (text/sources/query, answer/contexts), which can cause integration confusion. SKILL.md asks to npm install even though package.json contains no dependencies. The runtime instructions do not direct reading unrelated files, env vars, or network endpoints — they operate on provided inputs only.
Install Mechanism
No install spec in registry (instruction-only), but project includes package.json and suggests running npm install. package.json declares no dependencies; installing will not pull external packages beyond npm metadata. There are no downloads from untrusted URLs, no extract/install instructions writing additional binaries, and no obfuscated/install-time behavior.
Credentials
The skill requires no environment variables, no credentials, and the code does not read process.env or config paths. Requested privileges are proportional to its stated purpose.
Persistence & Privilege
always:false and no code that modifies agent/system configuration or other skills. The classes are in-process JS modules that emit events but do not persist credentials or alter global agent settings.
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 yuyonghao-evaluation-suite - 安装完成后，直接呼叫该 Skill 的名称或使用
/yuyonghao-evaluation-suite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
v0.1.2
Update - 2026-04-01
v1.0.1
Update from local workspace
v1.0.0
Initial release of yuyonghao-evaluation-suite.
- Provides evaluation suite for RAG, reasoning, and hallucination detection.
- Supports batch evaluation and unified API for multiple evaluation types.
- Includes modular evaluators: RAGEvaluator, ReasoningEvaluator, HallucinationDetector.
- Offers comprehensive documentation and quick start guides.
v0.3.0
- No changes detected in this version.
- The changelog remains the same as the previous version.
- Functionality, documentation, and code are unchanged from the prior release.
v0.2.0
Update to v0.2.0
v0.1.1
## v0.1.1 Changelog
- No file changes detected in this release.
- Documentation and functionality remain unchanged from v0.1.0.
v0.1.0
Initial release of the OpenClaw Evaluation Suite.
- Supports RAG evaluation (retrieval quality, answer relevance, context accuracy)
- Includes reasoning ability tests and hallucination detection for AI-generated content
- Provides batch evaluation and a unified evaluation API
- Modular design with dedicated classes for RAG, reasoning, and hallucination evaluation
- Simple configuration and usage via JavaScript imports
元数据
常见问题
Evaluation Suite 是什么？
Provides API for evaluating RAG quality, logical reasoning, and detecting hallucinations in AI-generated content with batch support. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 233 次。
如何安装 Evaluation Suite？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install yuyonghao-evaluation-suite」即可一键安装，无需额外配置。
Evaluation Suite 是免费的吗？
是的，Evaluation Suite 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Evaluation Suite 支持哪些平台？
Evaluation Suite 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 Evaluation Suite？
由 yuyonghao-123（@yuyonghao-123）开发并维护，当前版本 v0.1.2。
推荐 Skills