openagentprimitives

A secure way to run enterprise AI agents. Building blocks with per-action authorization, prompt-injection containment, least privilege, and a tamper-evident audit log.

GitHub

Official source

Start locally with the CLI guide