KATE is an open-source platform for tracing, evaluating, and improving AI agents. Auto-instrument your code, run quality evals, and visualize every decision your agent makes.
Drop-in SDK captures every LLM call, tool use, and chain step automatically. Zero config tracing.
Run faithfulness, relevance, and hallucination checks on every agent response. Powered by DeepEval.
Waterfall timelines show exactly how your agent thinks. Drill into any span to see inputs, outputs, and latency.
pip install kate-sdk
One-line init wraps your LLM calls automatically
Traces and eval scores appear in your dashboard instantly
Sign in with Google to get your API key and start sending traces in under 5 minutes.
Get Started Free