CRUX-in-a-box Open-Sources Infrastructure for Weeks-Long Agent Evaluations

Sayash Kapoor and the CRUX team released CRUX-in-a-box, an open-source infrastructure for running and monitoring weeks-long agent evaluations, distilled from six months of long-horizon evaluation practice, with features including VM-based long runs, risk flagging, and cost controls.

2026-11-06 ~ 2026-11-06 · 3 related posts

• CRUX-in-a-box: open-source infra for weeks-long, monitored agent evaluations — sayashk · 2026-11-06
• CRUX-in-a-box: open-source infra for weeks-long agent evaluations — sayashk · 2026-11-06

1 near-duplicate retellings: sayashk