[SCITT] Re: CCS (-09): a verifier-side runtime evidence record for agent tool calls - boundary with SCITT work (individual submission)

[Guigui Wang <[e-mail]>]

Konrad,

Thanks for reading -09 so carefully, and for stating the boundary in your own words. Your reading matches the intent: CCS is a bounded, verifier-side admission record for a single tool call, and RFC 9943 keeps payload content outside the SCITT charter. When a CCS receipt is carried inside a registered signed statement it is composition, not a dependency, and the draft says exactly that and requests no IANA allocations.

Two things in your proofbundle note are genuinely useful to us. First, your rule that a green run over accept-only cases is not conformance, and that every capability the corpus proves must also carry the rejection path, is the same anti-omission floor CCS tries to set for pre-execution verification; it is helpful to see it enforced in a shipping verifier rather than only asserted in text. Second, thank you for naming the open seam plainly - "provenance SHAPED, not provenance", with the gate verdict bound to the source tree and not yet to the shipped wheel. Collapsing that distinction is exactly the kind of grade inflation that makes evidence decay, and I would rather rely on partners who name the gap than partners who hide it.

Happy to keep the two efforts interoperable at the payload layer without coupling them.

Thanks,
Guigui Wang
Correctover - AI Reliability
draft: https://datatracker.ietf.org/doc/draft-correctover-ccs/
(individual submission, not an RFC or IETF endorsement)