Skip to main content
MCP schemas can vary by dataset type and operation. Do not guess a field name from a previous dataset or another operation.

​
For AI agents

  • Call get_dataset_fields for the current dataset_id.
  • Use the returned field names and types exactly.
  • Preserve the account and dataset identifiers used to retrieve the schema.
  • Explain when a requested field is unavailable instead of silently substituting one.

​
Discover fields

Call get_dataset_fields, then use its response to construct enrich_dataset, qualify_dataset, research_dataset, or transform_dataset arguments. See Supported MCP tools for the current schema.

​
Dynamic schemas

MCP field availability is dataset- and operation-specific, so the response from get_dataset_fields is the authoritative catalog for the current request. It may include field names, types, descriptions, and whether a field can be used for a particular operation. Use exact returned names in later calls. If a requested field is not present, ask the user to choose an available alternative rather than silently substituting one.

​
Field discovery before paid work

Field discovery is read-only. Inspect the schema before requesting a quote for enrichment, contact enrichment, qualification, or research. Preserve the same account_id and dataset_id on the quote, confirmation, and workflow polls. For advanced searches, request desired output columns in the natural-language query. The returned dataset schema determines the exact columns available afterward. See Advanced Dataset Creator in the MCP search reference.