safety-testing 开源项目推荐（6 个）

microsoft /PyRIT

微软开源 AI 红队测试工具包。通过自动扫描、自适应多轮攻击和 Web UI 测试生成式 AI 系统。

VenkateshDoijode /LLM-Strata

End-to-end Al/LLM security, safety and penetration testing framework for prompt injection, jailbreaks, RAG security, agentic attacks, red teaming, and monitoring.

TAKUYA0608-dev /agentic-star-agentic-ai-safety-testing-gate-agent

该项目在 GitHub 上没有填写描述

luizpaulo-antunes /llm-agent-evaluation-lab

Evaluation and observability lab for LLM agents: behavior, safety, latency, cost, redacted traces, and repeatable CI.

akalmaliky-spec /rulecheck-ar

Test LLM safety rules in Arabic: refusal accuracy and over-refusal across MSA and 6 dialects. YAML cases, a CLI, and a CI gate.

swatijreddy /deepeval-ai-support-assistant-eval

DeepEval LLM evaluation framework using Claude API — tests GEval, Hallucination, Answer Relevancy, Faithfulness, Contextual Precision/Recall, Bias, Toxicity, PII Leakage, and Tool Correctness on a mock AI customer support assistant.