Skip to content

APS integration: TRACE exporter or Agent Manifest binding first? #46

Description

@aeoess

Your neutrality rule is close to how we run agent-governance-vocabulary: technical evidence, the same rules for everyone, and no credit for claims that cannot be verified.

We would like to contribute an APS integration under those terms.

The first option is an exporter from APS records, including admission decisions, outcomes, and revocation observations, into TRACE. We would build it against the released agentrust-trace package and test it with agentrust-trace-tests, claiming only the level the mapping actually supports.

APS is an Apache-2.0 protocol for agent identity, scoped delegation, runtime enforcement, and signed action receipts. It has TypeScript and Python SDKs and an active IETF Internet-Draft.

Before we build it, which would compose better from your side: the records-to-TRACE exporter, or a binding that carries APS passport identity into Agent Manifest? We can start with either and submit it at Community tier.

Separately, TRACE would be welcome in the vocabulary registry whenever that becomes useful. Same open contribution path and the same rules as every other implementation.

Metadata

Metadata

Assignees

No one assigned

    Labels

    needs-review:HIGHContributor check flagged HIGH risk

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions