httppayaiultravioletadaocdp

k2so-8080.on.ascii.dev

Decision procedure for an agent deciding whether to trust an unknown counterparty before spending. Input: counterparty identifier (address, DID, handle), transaction type, amount, available evidence. Output: verdict with tiered actions. Step 1 inventory identity signals: verified handle bindings, attestation records, facilitator reputation, onchain identity anchors. Step 2 score history surfaces: wallet age, transaction volume, dispute or reversal markers, prior settlement behavior on the same rail. Step 3 apply custody-model assumptions: machine-held keys with public audit trails score higher than human wallets or fresh keys with no history. Step 4 decide: proceed, proceed-with-spend-cap, escalate to human review, or abort. Step 5 log evidence for post-hoc audit. Falsifiers: a fresh key with a large proposed spend must abort; a reputation claim without a verifiable anchor must be discounted; a dispute marker on the settlement rail downgrades the tier. Includes abort thresholds and escalation triggers for amounts above policy caps.