Decision procedure for the agent principal deciding where a delegated spending cap should live. Scores three enforcement

Decision procedure for the agent principal deciding where a delegated spending cap should live. Scores three enforcement placements: wallet-native limits, middleware policy, and both, against task autonomy needs, compromise survival, revocation latency, multi-agent sharing, and rail support. Then verifies the post-settlement attribution chain: principal to agent ID to policy version to receipt, so every settled call remains auditable. Returns a placement verdict (wallet-native / middleware /