AML Typology VoP Mismatch Hunt

Note (dated 2026-08-04): a chain composes independent computations performed in an analytical order. No step's output is carried into the next step by the execution model; each step computes from its own declared inputs. This is true of every chain on this site, including chains whose per-step copy below reads as a data handoff.

Reading mode: These runs can be gathered in any order; the sequence below is for convenience. Typology scoring and IBAN-name matching grade separate questions, and the session receipt closes the run set.

Score a synthetic transaction graph against AML typologies, then run a separate SEPA IBAN-name match batch over payee records the caller selects from the flagged accounts, and close with a session receipt. Each run computes from its own declared inputs, and each run hash is cited in the session receipt.

OpenChainGraph · 3 Steps compliance mandate Hash-Anchored §4 chain_depth:3 Client-Side · Zero PII

Chain Stages · 3 Steps across 3 Stages

1 ROOT node

AMLA Transaction Typology Risk Scorer art-10-amla-transaction-typology-risk-scorer

Typology risk scores per transaction with per-typology hit counts, plus flagged accounts and clusters from a synthetic transaction graph. Stage 1 of 3. Its execution hash is cited in the session receipt.

MCP Call · score_aml_typologies

{
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "name": "score_aml_typologies",
    "arguments": {}
  },
  "id": 1
}

↓

2 STAGE 2 node

VoP Batch Match-Rate Analyser art-11-vop-batch-match-rate-analyser

Batch IBAN-name match, close-match and no-match counts with the match rate at the declared strictness, over payee records the caller selects from the flagged accounts. Stage 2 of 3. Its execution hash is cited in the session receipt.

MCP Call · simulate_vop_matching

{
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "name": "simulate_vop_matching",
    "arguments": {}
  },
  "id": 1
}

↓

3 TERMINAL node

Agent-Action Audit-Trail Aggregator cry-05-agent-action-audit-trail-aggregator

Session receipt: one SHA-256 Merkle root over the ordered execution hashes of the runs above, with per-leaf inclusion proofs. Terminal stage. This step aggregates hashes and carries no compute proof of its own.

MCP Call · aggregate_execution_receipts

{
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "name": "aggregate_execution_receipts",
    "arguments": {}
  },
  "id": 1
}

How the runs relate

The caller maps flagged accounts to payee records. The flagged accounts and clusters from the first run are the caller's selection for the second run's batch. That mapping is caller-performed.

All inputs in this chain are synthetic or anonymised. Use synthetic transaction graphs and synthetic payee records. Scope: SEPA IBAN-name matching. US ACH account validation is out of scope for this chain.

Out of scope: US ACH account validation; screening of real accounts; any determination that a transaction is illicit or that a payee record is correct. This chain scores synthetic inputs and classifies name matches; it does not make a compliance determination.

The chain ends with the session receipt step, cry-05, as its terminal stage.

Question: A batch of SEPA payees is in front of us. Which records fall into the no-match band when the match threshold is raised to 0.95? Run the AINumbers MCP tool `run_chain` with {"chain":"aml-typology-vop-mismatch-hunt"} and read `art-11-vop-batch-match-rate-analyser.match_rate_pct`. Re-run with `inputs` for `art-11-vop-batch-match-rate-analyser`, reusing the values that result echoed, with `match_threshold` set to 0.95. Compare the same field. 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/aml-typology-vop-mismatch-hunt.html

AML Typology VoP Mismatch Hunt

Ask your agent