Agent Reliability Bench

Evaluate exact tool-call sequence compliance. This deterministic reference demo runs locally in your browser and does not send traces to a server.

Result

Run an evaluation to see results.

Exact sequence matching only. Not an LLM quality benchmark, model leaderboard, or semantic evaluation.

Source code and issues