*Need AI Agents for Testing*
A developer is seeking a few users who have their AI agent running with users to test a new observability and monitoring tool. The tool, built by the developer, tracks various issues related to AI agents, including hallucinations, prompt injection, bias, toxicity, and PII leaks.
*What the Tool Does*
The tool has several features, including:
* Tracking of various issues related to AI agents
* Prompt blocking
* Trace tree with token and cost calculation
* Integration with existing AI agents through two methods:
+ Proxy API (two-line change, suitable for no-code and quick integration)
+ SDK (full agent trace and observability)
*Why It Was Built*
The developer built this tool because they encountered the same problems when building AI agents themselves. Debugging LLM behavior is difficult and cumbersome, and existing tools either lacked the necessary features or were too complex.
*How to Try It*
If you have an AI agent running with users, you can try the new tool by commenting below or DMing the developer. They will share access and provide a quick setup (around 5 minutes). The testing is free, and anyone who wants to continue using the tool will be upgraded to the Pro plan for lifetime.
It's worth noting that the developer is looking for users who have their AI agent running with users, implying that this tool is designed for developers who are already working with AI agents and want to improve their observability and monitoring capabilities.