Two-step annual attestation cycle over a declared architecture type and component inventory. Step 1 derives the applicable mandatory and advisory control set from the policy-supplied control-framework version, scores declared implementation status, and returns coverage figures, a gap list keyed by published control number, an evidence index aligned to the framework's own supporting-evidence column, and an explicit not-applicable set with stated reasons so an omission cannot read as a pass. Step 2 tests assessor eligibility: assessment route, claimed certifications against the required set, independence from the implementer by distinct-identity counting, and assessment-date validity against the attestation deadline. Step 1 is gated hold while any mandatory control gap remains. Step 2 is gated reject, a terminal human rejection, when the assessor is ineligible, and the failing predicate is carried as the reason code. Release of the attestation bundle is a dual-control act with the assessor bound as reviewer and the attesting officer as approver; an examiner role binding grants inspection only and never an approval path. The framework's annual 31 December attestation deadline is a fact about the regime. This surface produces the firm's own evidence: it is not a Swift endorsement, not an assessor accreditation, and does not satisfy a KYC-SA submission.
hold until the gap is closed or explicitly dispositioned.check_cscf_control_applicability{
"jsonrpc": "2.0",
"method": "tools/call",
"params": {
"name": "check_cscf_control_applicability",
"arguments": {}
},
"id": 1
}
reject, carrying the failing predicate as the reason code. Attestation release is a dual-control act with the assessor bound as reviewer and the attesting officer as approver; an examiner role binding grants inspection only, never an approval path.check_assessor_independence{
"jsonrpc": "2.0",
"method": "tools/call",
"params": {
"name": "check_assessor_independence",
"arguments": {}
},
"id": 1
}
Copy this paragraph into Claude, OpenClaw, or any MCP-aware agent to run this whole workflow on the server and check one field that answers a real question.
Question: We run a Swift architecture A1 setup. How much of the mandatory control set is covered by evidence we already hold?
Run the AINumbers MCP tool `run_chain` with {"chain":"swift-csp-attestation-cycle"} and read `art-486-cscf-control-applicability.mandatory_coverage_pct`.
Verify: call `verify_execution_hash` (https://mcp.ainumbers.co/mcp) with `claimed_hash` set to `composite_execution_hash` and the full `composite_artifact`.
Ledger, for a human re-check: https://ledger.ainumbers.co/
PII rule: send synthetic or anonymised inputs only. The MCP server runs these kernels and logs no payloads.
Chain page: https://ainumbers.co/chaingraph/chains/swift-csp-attestation-cycle.html