Built for the age of AI agents.
PandaProbe is the open-source stack for self-repairing agents, built by Chirpz AI. We give teams the tracing, evaluation, and repair infrastructure their agents need to learn and recover from failures on their own.
Making AI agents production-ready.
AI agents are fundamentally different from traditional software. They fail in unexpected ways, at unexpected times. Agent are also fundamentally different from LLMs in terms of metrics, and require a new set of tools for successfull evaluation and monitoring. PandaProbe exists to bring the same level of engineering rigour to AI agents that modern teams already apply to their conventional software services.
We started building PandaProbe because we couldn't find a tool that treated agent observability as a first-class concern. So we built one — and decided to make it open source from day one.
Open by default
Everything we build starts open source. Transparency is not a feature — it is the foundation.
Built for builders
PandaProbe is designed by engineers who ship AI agents. We solve the problems we live every day.
Backed by research
As a team of PhD researchers specialising in uncertainty and robustness in AI agents, every solution we ship is grounded in science.
Community first
The best ideas come from the people using the product. We build in public and ship based on real feedback.
We're building in the open.
Star the repo, open an issue, or just say hi on GitHub.