Start with env args like AGENT_ID for indicating which Merkle hash of which model(s) generated which code with which agent(s) and add those attributes to signed (-S) commit messages. For traceability; to find other faulty code generated by the same model and determine whether an agent or a human introduced the fault.
Then, `git notes` is better for signature metadata because it doesn't change the commit hash to add signatures for the commit.
And then, you'd need to run a local Rekor log to use Sigstore attestations on every commit.
All contributors must indicate in the commit message of their contribution if they used AI to create them and the contributor is fully responsible for the content that they submit.
This can be a label such as Assisted By: <Tool> or Generated by: <Tool> based on what was used.
But which metadata is better stored in git notes than in a commit message? JSON-LD can be integrated with JSON-LD SBOM metadata
Comments
From https://news.ycombinator.com/item?id=46660151 :
Does Entire solve for this?
Re: AI and DevOps Traceability: https://gemini.google.com/share/4c0c79c0f136
From https://github.com/pulp for example:
But which metadata is better stored in git notes than in a commit message? JSON-LD can be integrated with JSON-LD SBOM metadata
I made this last week in my spare time, without $60m in seed funding!
https://github.com/re-cinq/claudit