Authenticate to the RMS Risk Modeler API and confirm access to the required peril models (e.g., North Atlantic Hurricane, US Earthquake)
Format the exposure data as a location data file following the RMS EDM schema: location ID, address, geocoords, construction class, occupancy, TIV by coverage
Upload the exposure file via the API's location import endpoint and wait for the ingestion job to complete
Initiate a loss analysis run specifying the model, peril, financial perspective (gross, net of reinsurance), and return period set
Poll the analysis job status endpoint until the run completes, then retrieve the event loss table (ELT) and occurrence exceedance probability (OEP) curve
Download the results and load them into the reinsurance pricing model or treaty negotiation workflow
Known gotchas
Catastrophe model runs can take minutes to hours depending on portfolio size; design the workflow with asynchronous job polling rather than synchronous waits
Model version pinning is critical for reinsurance treaty comparisons; ensure both cedant and reinsurer are using the same model version when comparing loss estimates
Geocoding quality directly impacts modeled losses; a significant percentage of locations with low geocoding confidence should trigger a data quality review before pricing
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