Four-stage Know-Your-Agent pipeline that intercepts an agentic transaction and gates it through: prompt-injection detection, FATF customer risk scoring, Visa TAP cryptographic identity verification, and AML mandate compilation. Any stage failure blocks the chain.
Scan the inbound agentic transaction context: the tool call text, mandate payload, or agent instruction: for prompt-injection and tool-poisoning smells. A risk score ≥ 60 blocks the pipeline.
Paste the JSON output from T282, or enter the risk score (0–100) manually.
Score the customer or counterparty across 6 FATF Rec 10/12 dimensions. Risk tier High or Prohibited blocks the pipeline. Medium requires enhanced monitoring note in the final mandate.
Paste the Policy Mandate JSON from T477, or enter the risk tier manually: Low / Medium / High / Prohibited.
Inspect the Visa TAP HTTP Message Signature to cryptographically verify the agent's identity and TAP readiness score. A readiness score below 60 or a signature failure blocks the pipeline.
Paste the JSON output from T286, or enter readiness score (0–100) manually.
Compile the final AML Policy Mandate JSON incorporating the gating decisions from stages 1–3. The mandate encodes agent_instructions, risk tier, TAP status, and an immutable audit chain for downstream payment rails.
Paste the Policy Mandate JSON from T131. This becomes the final KYA clearance record.