Research and enrichment
Build a new list, enrich existing sheet rows or submit typed research in batches.
Turn your target-market criteria into a contact-ready list. Research the facts that decide account fit, then add work emails and direct phone numbers to the people you want to reach. Choose a new list, an existing Sheet or a batch of answers for your application.
A new-list request defines its criteria, target count, identity field and typed result fields. Leadbase creates a private sheet and saves verified matches as rows. Open it with view_sheet in a compatible MCP client, or follow its Leadbase link. Track sheet progress with sheets_research_status and job results and billing with enrichment_get.
For an existing sheet, choose a configured column or define a new column with a question and result type. Supply specific row IDs for a subset. Omit them only when you want all existing rows, up to 10,000. sheets_enrich skips populated and active cells. Read sheets_get_enrichment_status and sheets_query to inspect the result. Explicit replacement and multi-column targets use sheets_start_enrichment and its current schema.
Without a sheet, submit known facts, a question and an output type per item. You can request email addresses or phone numbers for many people in one batch. See typed enrichment, contact details and the bulk example.
Direct HTTP operations
For an application that already knows the exact filters or provider parameters, the HTTP API also supports structured company and people search and direct catalog execution with tools.list and tools.run. These operations are not exposed as public MCP tools. MCP agents use the research tools above and let Leadbase select the underlying sources.
Give research a clear target
Provide identifiers such as a domain or professional profile URL. Ask a focused question per item and choose the narrowest useful result type. A search candidate is not a verified match for every criterion; inspect identity and supporting sources.
Start with a representative sample and inspect missing results and actual cost before expanding. Accepted jobs run independently of the browser or client. Keep their IDs and original retry keys. An empty result, an execution failure and pending billing are different outcomes. See Activity and Credits.