Evaluation Suite
v0.1.2Provides API for evaluating RAG quality, logical reasoning, and detecting hallucinations in AI-generated content with batch support.
⭐ 0· 114·0 current·0 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & 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.
Assessment
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.Like a lobster shell, security has layers — review code before you run it.
latestvk97dr54czv0751v8y5wmtk6cp18414mh
License
MIT-0
Free to use, modify, and redistribute. No attribution required.