PandaProbe is an open-source agent engineering platform built to give developers visibility into how AI agents behave, tracking every tool call, decision, and model interaction an agent makes. It comes from Chirpz AI, an applied AI lab founded in 2025 by Sina Tayebati and based in Chicago, and is released under an Apache 2.0 license.
The platform groups every trace into a session representing an agent's full lifecycle, so even when execution fans out across parallel or asynchronous subagent calls, developers can reconstruct the complete execution tree in one view. Evaluation metrics score how an agent performs across a full workflow rather than a single response, and regression detection flags when a new deployment makes an agent less reliable, with SDKs that plug into LangGraph, CrewAI, and the Claude Agent SDK.
PandaProbe fits engineering teams shipping AI agents to production who need debugging and eval tooling beyond print statements and manual review. The self-hosted, open-source version is free; PandaProbe Cloud adds a managed Pro plan starting at $29 per month for small teams, plus an Enterprise tier with SSO and custom hosting.
