Collect the full set of reason codes and their definitions from each IDV vendor's documentation and maintain an internal mapping to standardized internal rejection categories
In your decision processing logic, inspect both the overall decision status and the array of reason codes or sub-results returned by the vendor
Classify each reason code as auto-reject (for example, suspected document fraud), manual review required (for example, data extraction failure or unclear image), or customer-actionable retry (for example, image too dark)
For manual review cases, persist the full IDV result payload, the reason codes, the applicant PII, and any available document images in a case management system accessible to your review team
Implement SLA tracking on the manual review queue to ensure cases are reviewed within your regulatory or internal time commitment
Feed final manual review outcomes back to the vendor if the vendor supports reviewer feedback loops for model improvement
Known gotchas
Reason code definitions and values are vendor-specific and version-specific; after a vendor SDK or API update, re-validate your reason code mapping table before deploying
Auto-rejecting on every non-passing reason code without human review can cause discriminatory outcomes for certain document types or populations; audit rejection rates by demographic segment
Manual review queues require trained staff with access to sensitive PII and biometric data; ensure your access control, data handling, and audit logging meet your compliance requirements
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