Pre-action verification for AI agents: a signed verdict before the action runs, independently checkable, not a log you have to trust. Also co-authoring a couple of ERC standards for agent identity/reputation. api.babyblueviper.com