Run a list-building and research agent

Run a list-building and research agent

Starts an Exa Agent run that searches and reads the web to build lists, enrich records you pass in or research a question, and returns cited findings with optional schema-validated JSON.

Note: Runs asynchronously: poll GET /agent/runs/{id}, or on MCP pass the returned id back as runId while status is running. effort trades cost for completeness, and budget caps what an auto or ultra run can spend.

Set up

Use the first option your agent supports.

MCP (official, remote)

Add this server to your agent's MCP settings, then sign in when asked.

{ "mcpServers": { "exa": { "url": "https://mcp.exa.ai/mcp?login" } } }

Call the MCP tool agent_run.

Server URL: https://mcp.exa.ai/mcp?login

API (official)

• Base URL: https://api.exa.ai
• Endpoint: POST /agent/runs
• Auth: send the header Authorization: Bearer $EXA_API_KEY
• Get a key: https://dashboard.exa.ai/api-keys
• Docs: https://exa.ai/docs/get-started/quickstart

Before anything else, confirm access with the cheapest read-only call, like a list or a search. Never send or change anything to test access.

Rules

• Ask the user before anything that sends messages, costs money, or changes data, and say how many records it touches. One approval covers a batch the user has seen.
• Never print API keys.