The agent produces a cryptographically signed JSON artifact containing a customer's risk tier and compliance flags, with a SHA-256 hash that allows third parties to verify the decision without trusting the intermediary.
The agent takes customer risk profile parameters (e.g., PEP status, sanctions hits, jurisdiction risk) as input. It runs a deterministic kernel to calculate a risk score and tier. It then constructs an OpenChainGraph v0.4 artifact envelope, including the required fields like tool_id, mandate_type, and policy_parameters. Crucially, it computes the execution_hash by canonicalizing the inputs and outputs using RFC 8785 and hashing them with SHA-256. The agent returns the full artifact, enabling client-side verification of the compliance decision.
- Category: Finance and payments
- Effect: Changes data in your accounts
- MCP servers: AINumbers MCP (https://koskamo.com/explore/ainumbers-mcp)
- Source: OpenChainGraph Integration Guide (https://ainumbers.co/chaingraph/ocg-integration-guide.html)
- Source quote: "Every conformant tool takes a policy_parameters object as input, runs a deterministic pure-function kernel, and emits a JSON artifact that includes a cryptographic execution_hash committing to both the inputs and the output."
- Playbook page: https://koskamo.com/explore/use-cases/generate-verifiable-kyc-compliance-artifacts
Prompt: Generate a verifiable OpenChainGraph v0.4 artifact for a KYC check. Input parameters: entity_name='ACME Corp', pep=false, sanctions_hit=false, high_risk_jurisdiction=true. Use tool_id 'my-kyc-checker' and mandate_type 'compliance_mandate'. Calculate the risk score (30 for PEP, 50 for sanctions, 20 for high risk jurisdiction) and assign tier: high if score >= 50, medium if >= 20, else low. Compute the SHA-256 execution hash of the canonical JSON preimage of policy_parameters and output_payload. Return the full JSON artifact.