Agentic QA - 开源中间件，用于对循环中的代理进行模糊测试
我之所以构建这个，是因为我看到我的 LangChain 代理在一夜之间因无限循环而消耗了大约 50 美元的 OpenAI 额度。
这是一个中间件 API，充当“飞行模拟器”。你可以将代理的提示发送给它，它会进行对抗性攻击（红队测试），以在部署之前捕捉循环和个人信息泄露。
代码和仓库： [https://github.com/Saurabh0377/agentic-qa-api](https://github.com/Saurabh0377/agentic-qa-api)
在线演示： [https://agentic-qa-engine.onrender.com/docs](https://agentic-qa-engine.onrender.com/docs)
欢迎反馈你们见过的其他失败模式！
查看原文
I built this because I watched my LangChain agent burn ~$50 in OpenAI credits overnight due to an infinite loop.<p>It's a middleware API that acts as a 'Flight Simulator'. You send it your agent's prompt, and it runs adversarial attacks (Red Teaming) to catch loops and PII leaks before deployment.<p>Code & Repo: https://github.com/Saurabh0377/agentic-qa-api
Live Demo: https://agentic-qa-engine.onrender.com/docs<p>Would love feedback on other failure modes you've seen!