CHRONOS alignment follow-on: method-family + non-authority field on proof-claim contract (per sociosphere#498)

[mdheller]

Context

SocioProphet/socioprophet#498 tracks the overall KAIROS↔CHRONOS alignment thread. sociosphere/docs/integration/neurosymbolic-chronos-alignment.md's "Definition of done" names Holmes as one of the repos needing a downstream follow-on item (Holmes isn't in the doc's authority-plane table itself, but is explicitly named in the DoD list). Repo name confirmed via gh repo list SocioProphet | grep -i holmes: SocioProphet/holmes ("Holmes Language Intelligence Fabric") — there is no separate holmes-search repo; sherlock-search is a distinct, related repo.

What already exists in this repo

Holmes already has a reasoning-layer contract that maps almost directly onto CHRONOS's method-family table. Per docs/PROOF_CLAIM_CONTRACT.md, Holmes owns Propose -> Explain -> Verify in the canonical loop Observe -> Anchor -> Normalize -> Propose -> Explain -> Verify -> Govern -> Act -> Receipt -> Learn, with the contract invariant VectorCandidate | ModelOutput | GraphCandidate -> ProposedClaim -> ExplanationTrace + ProofCertificate? + ContradictionReport. It defines a TruthBounds object mapped to ontogenesis.socioprophet.dev/v1/TruthBounds with a ReasoningTrace format carrying ruleName, premises, conclusion, evidenceRefs, confidence, and truthBounds (lower, upper, method) — this is functionally the same shape as the alignment doc's "LNN-style truth-bound propagation: advisory bound/inconsistency analysis" row. Holmes explicitly states it "does not admit claims into policy-approved truth" — Policy Fabric decides allow | deny | require_review | provisional.

The concrete gap

Holmes' ProposedClaim/ExplanationTrace/ProofCertificate/ContradictionReport chain already produces almost every field the alignment doc's carrier-boundary section requires (source evidence reference, explanation trace, validation-adjacent state) — but it doesn't currently declare a method-family tag distinguishing, say, an LNN-style truth-bound claim from a NeurASP-style adjudicated claim or a plain vector-candidate-derived claim, nor an explicit "owning authority plane" / "non-authority declaration" field on the claim package itself (ownership is described in prose in PROOF_CLAIM_CONTRACT.md, not carried as data on the artifact).

Ask

Look at whether the ProposedClaim / ExplanationTrace / TruthBounds contract (docs/PROOF_CLAIM_CONTRACT.md, examples/holmes-proof-claim-contract.json) should carry a method-family field and an explicit non-authority declaration, so the claim package can state (as data, not just doctrine) which of the alignment doc's method families it was produced under and that Policy Fabric — not Holmes — holds admission authority. This is a request to look at the connection, not a mandate to change the existing contract invariant.

Non-goals

• Holmes keeps sole ownership of propose/explain/verify artifacts; this does not ask Holmes to gain policy admission authority — the contract already states Holmes does not admit claims into policy-approved truth.
• No canonical-schema authority moves into sociosphere; any new field would still map through Ontogenesis's canonical contracts as the existing ones do.

Refs: SocioProphet/socioprophet#498, sociosphere/docs/integration/neurosymbolic-chronos-alignment.md, docs/PROOF_CLAIM_CONTRACT.md (this repo).