Description
When an AI agent fails, it's hard to tell which reasoning step or tool call went wrong, or whether a fix helped. PandaProbe is an open source agent engineering platform with traces, evals and metrics to debug and improve AI agents.
It integrates LangGraph, CrewAI, the Claude Agent SDK, OpenAI Agents SDK and more, self-hostable.
Traces:Every step.
Evals:Compare results.
Metrics:Quantified gains.
Frameworks:LangGraph, CrewAI and more.
It integrates LangGraph, CrewAI, the Claude Agent SDK, OpenAI Agents SDK and more, self-hostable.
Features
Traces:Every step.
Evals:Compare results.
Metrics:Quantified gains.
Frameworks:LangGraph, CrewAI and more.
