Last released Jul 29, 2026
Eval/replay harness for AI agents: run agent tasks in isolated E2B sandboxes, record full traces, measure pass rate, flakiness and cost.
Supported by
