Inspect

Framework / harness (not an eval — tooling/reference entry)

De facto standard eval framework for frontier-model and agent evals; sandboxes, tool use, scoring.

Category: Eval frameworks & harnesses
Insurance exposure: All
Publisher: UK AI Security Institute (now hosted with Meridian Labs)
Publisher type: Government / standards body
License: MIT
Status: Active (<90d)
Last commit: 2026-10-15
Stars: 3014
Notes: Used by METR, Anthropic, labs and AISIs. Mirror at meridianlabs-ai/inspect_ai.

Wanich Risk Lab, 2026.