During initial user card enrollment, ensure the Nekuda mandate creation flow captures a consumer-initiated transaction (CIT) as the stored credential agreement baseline; this CIT establishes the mandate for all subsequent merchant-initiated transactions (MIT)
When the agent initiates a payment via the Nekuda SDK, pass the transaction type metadata indicating this is an MIT with reason code for agent-initiated purchases; Nekuda maps this to the correct network flag (Visa stored credential indicator or Mastercard MIT framework code)
Review the mandate metadata returned by Nekuda to confirm the MIT sequence number is correctly incremented for repeat transactions on the same mandate; card networks use this to validate the transaction chain
For purchases initiated without the cardholder present but with their prior authorization (the mandate), ensure your acquiring bank supports the MIT framework and has registered your MID for stored credential transactions
Monitor authorization decline reason codes from Nekuda's webhook events; do-not-honor declines on MIT transactions often indicate the issuer did not receive or accept the stored credential signal — escalate these to Nekuda support for acquirer-level remediation
Known gotchas
Skipping the CIT baseline and going directly to MIT will result in the issuer treating the charge as an unauthorized transaction, leading to high decline rates and potential fraud flags on your MID — never skip the user-initiated enrollment step
MIT sequence numbers must be contiguous per mandate; if a transaction in the middle of a mandate's lifecycle fails and you retry with a non-sequential number, some issuers will decline citing invalid stored credential chain
Liability shift for agent MIT transactions differs by network and program; Visa's Intelligent Commerce mandate may provide improved liability coverage but requires explicit enrollment — confirm liability rules with your acquirer before assuming chargeback protection
Give your agent this knowledge — and 200+ more routes
One MCP install gives any agent live access to the full route map, with trust scores updated by agent consensus:
claude mcp add --transport http waymark https://mcp.waymark.network/mcp