Register an AI agent with Visa's Agentic Directory and implement Agent Score consumption in your merchant fraud stack

Verified steps

1. Apply for Visa Intelligent Commerce (VIC) sandbox access through the Visa developer portal; once approved, complete the Agentic Directory registration for each agent or agent platform you operate — the Directory is Visa's verification layer that confirms agents and merchants are legitimate participants before routing live transactions
2. Integrate the Visa MCP server into your development environment; the MCP server exposes VIC API capabilities as plain-language tool calls compatible with Claude, Gemini, and other MCP-compatible agent hosts, reducing the integration effort versus direct REST calls
3. At transaction time, consume the Agent Score signal returned in the authorization response or via the VIC pre-auth API; Agent Score is a risk score from Visa (co-developed with New Generation) that reflects agentic commerce readiness and transaction risk — wire it into your fraud decision engine as a feature alongside your existing signals
4. Implement differentiated fraud rules for agent-identified transactions: velocity rules tuned for human shoppers will generate false positives on legitimate agent patterns (repeated purchases, off-hours activity) — create a separate rule set for transactions where an Agentic Directory entry confirms the agent identity
5. Use Intelligent Commerce Connect as your single integration point for multi-protocol agent acceptance; the product routes agent-initiated transactions from Trusted Agent Protocol, Machine Payments Protocol, ACP, and UCP through one merchant integration rather than four separate ones
6. Complete VIC certification before enabling live agent traffic; the certification process validates your mandate parsing, revocation handling, and Agent Score consumption against Visa's test scenarios