dingo data quality
v1.0.4Evaluate AI training and RAG data quality using rule-based or LLM-based metrics with Dingo's flexible, multi-format assessment framework and CLI/SDK support.
⭐ 3· 142·0 current·0 all-time
bychupei@e06084
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
Name/description (data quality / RAG evaluation) match the SKILL.md, _meta.json, and the included fact_check.py script. The package asks for an LLM API key only for LLM-based evaluators or the ArticleFactChecker flow, which is appropriate for the stated functionality.
Instruction Scope
SKILL.md instructs using local dataset files and CLI/SDK calls and shows configs that reference only the expected inputs (data files, evaluator configs, optional API keys and endpoints). The script reads input articles and writes temporary JSONL and output artifacts; it explicitly blocks special system paths and symlinks. No instructions attempt to read unrelated system secrets or send data to unexpected endpoints beyond the LLM/search APIs you configure.
Install Mechanism
This is instruction-only (no install spec). SKILL.md recommends installing the published dingo-python package via pip (a standard registry install). No downloads from arbitrary URLs or archives are present in the skill bundle itself.
Credentials
LLM-based evaluation requires an OpenAI-compatible API key (OPENAI_API_KEY) and optionally OPENAI_BASE_URL, OPENAI_MODEL, and TAVILY_API_KEY for web search — all proportional to an LLM-driven fact-checker. Note: registry 'required env vars' field is empty but the code and docs clearly treat OPENAI_API_KEY as required for LLM flows; this is expected but worth confirming before enabling LLM mode.
Persistence & Privilege
Skill does not request always: true, does not modify other skills or global settings, and is not installing persistent background agents. Autonomous invocation is allowed (default) but not combined with other red flags.
Assessment
This skill appears to do what it says: rule-based checks run offline without keys, while LLM-based/fact-checking requires an OpenAI-compatible API key (OPENAI_API_KEY) and optionally a search API key (TAVILY_API_KEY). Before using LLM mode: (1) confirm you trust the dingo-python package source on PyPI or the linked GitHub repo, (2) provide API keys only for providers you intend to bill, (3) be aware the tool will send article text and extracted claims to whatever API_BASE_URL you configure, so verify that endpoint, and (4) you can use rule-based evaluators without giving any credentials. If you want extra assurance, inspect the upstream dingo-python package code and confirm network behavior and telemetry before supplying secrets.Like a lobster shell, security has layers — review code before you run it.
latestvk97f29by9zyj3q24astdb0ygbx83hrg5
License
MIT-0
Free to use, modify, and redistribute. No attribution required.