Skip to content

Data Science Agent

Ask Runs GitHub ↗

Checking…

Ask Runs Reports Datasets

About this system

Benchmarks Evaluations Failures Research Compare runs Compare datasets MCP

Benchmarks, evaluations and audit surfaces -- measurements of the agent, not of your data.

On this machine

Settings that belong to the instance you are talking to, not to the analysis.

Evidence before claim · Ask. Analyze. Verify. Reproduce.

GitHub Docs API /health /version

Step 3 · Compare runs

Compare runs

Side-by-side evidence output for two runs. Emerald marks the larger value.