Tag: sandbox

Claw-Eval is an open-source evaluation harness for LLM agents featuring 300 human-verified tasks, 2,159 rubrics, and a strict Pass^3 consistency protocol. Maintained by PKU and HKU researchers.

Discover nono, the kernel-isolated sandbox from Sigstore's creator. Zero-latency capability-based security for AI agents with credential injection, snapshots, and cross-platform support. Stop hoping your agents won't be compromised—guarantee they can't be.