Runx

Runx is an open-source, governed runtime that sits beneath AI agents, executing portable agent skills under explicit authority, delivering credentials without exposing them to the model, and sealing every run into a verifiable receipt that proves what ran, who authorized it, and what happened. It works alongside whatever agent framework or orchestration layer a team already uses, with a local CLI as the primary surface and an optional hosted skill catalog. It's for teams running AI agents on real, consequential work who need auditable proof of what an agent actually did instead of trusting its own account of it.