[WIMSE] New version: draft-bu-agentproto-security-principal-binding-03

[Songbo Bu <[e-mail]>]

Hello all,

I have submitted draft-bu-agentproto-security-principal-binding-03:

https://datatracker.ietf.org/doc/draft-bu-agentproto-security-principal-binding/

This revision makes the verifier-facing model more concrete. The main
changes are:

   - a scoped-standing-grant row (C-013), with the grant semantics
   separated from opaque carriers and per-action authorization;
   - constrained patterns for condition-bound possession and optional
   human-principal assertions;
   - explicit separation of wire freshness, status freshness,
   condition-liveness, clone-detection signals, and established-session
   behavior;
   - byte-exact rules for digest-based composition and additional negative
   vectors for representation mismatch, unverified cryptographic fields,
   stream framing, emulated roots, and session resume; and
   - tighter requirements for claiming implementation status and evidence
   scope.

I would especially value focused review of the following points:

   1. whether C-011 and C-013 now keep claims, carriers, verifier roles,
   and constrained results sufficiently distinct;
   2. whether the condition-bound row draws the right boundaries among
   measured enrollment inputs, transcript freshness, condition-liveness, clone
   signals, and established sessions;
   3. whether the digest-composition rules are precise enough for
   interoperable implementations and useful negative tests; and
   4. whether the optional human-principal row avoids conflating
   association, consent, authorization, and accountability.

Thank you to Leonard Gebauer, Chris Hood, Iman Schrock, Steven Mih, Tom
Sato, Anton Sokolov, Thi Nguyen-Huu, Sergei Nikitin, John O’Leary, Karthik
Rampalli, Akira Okutomi, Blake Morrison, and the AGENTPROTO, WIMSE, and
SCITT list participants whose discussions helped sharpen these boundaries.
The draft’s acknowledgments identify the corresponding contribution areas
in more detail.

Best,

Songbo