Connecting an AI Agent

The built-in chat on the Explore page handles most research needs. Connecting an external AI agent gives you additional power: scripted workflows, batch operations, and integration with your existing development tools.

Claude Code

Copy the command from the Connect page in your dashboard:

claude mcp add --transport http lomita https://YOUR-INSTANCE-mcp.lomita.io/mcp

Your browser opens for sign-in — no API key copy-paste needed. Start a new Claude Code session and Lomita tools are available.

Claude Desktop

Download the pre-filled config from the Connect page, or add manually:

Settings > Developer > Edit Config:

{
  "mcpServers": {
    "lomita": {
      "type": "http",
      "url": "https://YOUR-INSTANCE-mcp.lomita.io/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}

Get your API key from Settings in the dashboard.

OpenCode / Other MCP frameworks

Use the JSON config from the Connect page. Any framework that supports HTTP MCP transport works:

{
  "type": "http",
  "url": "https://YOUR-INSTANCE-mcp.lomita.io/mcp",
  "headers": {
    "Authorization": "Bearer YOUR_API_KEY"
  }
}

What your agent can do

Once connected, your agent has access to these tools:

Tool	What it does
`research`	Ask a question — agents find data and produce a report
`follow_up`	Dig deeper into a previous finding
`status`	Check progress on active research
`sources`	Search available data sources
`connect_integration`	Connect your own API, CRM, or database
`deliver`	Send a report to email, Slack, or webhook
`monitor`	Set up continuous monitoring with alerts
`upload`	Add your own CSV, JSON, or Parquet data

No LLM setup required

The research agent team uses a bundled AI model. You don't need to configure a provider — research works immediately.

If you want to use your own model (Anthropic, OpenAI, Gemini), you can optionally override with set_provider. This is not required.