Generate a Monte Carlo API key from Settings > API in the Monte Carlo UI; the key consists of a keyId and keySecret used as HTTP headers x-mcd-id and x-mcd-token respectively
POST GraphQL queries to https://api.getmontecarlo.com/graphql; use the introspection query to explore available types and operations, or reference the Monte Carlo API explorer at docs.getmontecarlo.com/graphql
To fetch table lineage, send the query getTableLineage(mcon: 'YOUR_TABLE_MCON', direction: 'downstream', hops: 2) which returns downstream tables up to 2 hops away; MCON (Monte Carlo Object Name) is the unique identifier for tables in the Monte Carlo graph
To create an incident programmatically, use the mutation createOrUpdateIncident with fields for tableId, incidentType, severity, and a description; this is useful for injecting externally detected anomalies into Monte Carlo's incident workflow
Use getAlerts to retrieve active alerts and getIncidents to list open incidents; both accept filter parameters for table MCON, time range, and severity to scope results
Known gotchas
MCONs are Monte Carlo's internal identifiers for data assets; they are not the same as table names or fully qualified names — look up the MCON for a table using the search query before using it in lineage or incident mutations
Monte Carlo's GraphQL schema is updated continuously; field names and required arguments may change between releases — pin your queries against the API explorer version documented in your onboarding materials and re-test after Monte Carlo updates
Creating incidents via API bypasses Monte Carlo's anomaly detection thresholds; incidents created this way are marked as manual and do not affect the automated anomaly scoring models
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