Receipts for agent-rendered videos in hyperframes?

[hheskihoran]

hyperframes' pitch, "Write HTML. Render video. Built for agents.", made me think about provenance: when one agent asks another to render, how does the caller know the render happened, beyond the renderer's own report?

I maintain INAM Protocol (disclosure), an open protocol where both parties sign an Execution Receipt. A receipt could bind the input HTML hash to the output video hash, and a verifier can add a verdict. It proves the render occurred, not that the video is good.

Would you want render provenance in the core, or kept as an external plugin? Repo: https://github.com/inamprotocol

[capitaharlock]

For render provenance, I’d put a protocol-neutral render-completion hook in hyperframes core and keep receipt generation and signing in an external plugin. That gives the plugin a place to attach without making core depend on INAM’s receipt format.

[hheskihoran]

Agreed. A protocol-neutral render-completion hook in core, with receipt generation and signing in an external plugin, is the better split: core stays free of any receipt format, and INAM becomes one plugin among possible others.

What the hook would need to expose for any provenance plugin: the composition/spec input (or its hash), the output file path or hash, start and finish timestamps, and the renderer version. If maintainers are open to a hook like that, I'm happy to draft it as a small PR and keep the INAM plugin in a separate repo.