You open one tool to build the list, another to find emails, another to verify phone numbers and another for company data. Four logins, four billing dashboards, and four places for a workflow to break.

The Databar MCP server puts all of it behind one connection. Connect Claude, Cursor or any other MCP client once, and your AI assistant can search and run enrichments across 100+ data sources, run waterfalls, build and update tables, and push results to your CRM, all from plain-English requests like "find the email for John Smith at Google using a waterfall".

We launched it on Day 1 of Launch Week, alongside the Databar API, Python SDK and CLI. This post covers what the MCP server does, how to connect it, and when to use it instead of the API.

What the Databar MCP server gives your agent

The server exposes tools that map to the public Databar REST API, grouped into six areas:

• Enrichments. Search the catalog by keyword or category, check an enrichment's inputs, outputs and price, then run it on one record or in bulk.

• Waterfalls. Run a waterfall enrichment that tries several providers in order until one returns a result, with optional email verification. Waterfalls can run on their own or be attached to a table.

• Tables and columns. Create, rename and delete tables and columns, attach enrichments or waterfalls to a table, and run them on all rows, empty rows or rows that errored.

• Rows and folders. Insert, patch, upsert, filter and delete rows, and keep tables organized in folders.

• Exporters. Attach a CRM or other destination to a table, map the fields and trigger the export.

• Account. Check your credit balance, with built-in spending guardrails.

The full, current list is in the MCP tools reference.

What this looks like in practice

Here's a campaign build that used to mean several tools, CSV exports and some glue code. With the MCP server connected, it's a handful of requests in one conversation:

1. "Create a table called Series A SaaS and add these 40 company domains."

2. "Enrich each company with headcount, industry and latest funding."

3. "Find the VP of Sales at each company, then get their work email with a waterfall."

4. "Show me the rows where the email came back empty, and try the phone waterfall on those."

5. "Push the finished rows to HubSpot."

Your assistant picks the right enrichments, maps the columns, runs the jobs and reads the results back, and everything lands in a Databar table you can open and check. For a full walkthrough, see how to build an outbound campaign with Claude Code.

A few more prompts to try once you're connected:

• "Verify these 50 emails and flag any that fail before I add them to the sequence."

• "Find the tech stack for these 10 competitor domains and summarize which marketing tools they use."

• "List my tables and show the columns for the first one."

• "How many credits do I have left?"

How to connect it

Use the hosted server. It gets new tools and fixes first, and there's nothing to install or update:

https://mcp.databar.ai/mcp

https://mcp.databar.ai/mcp

https://mcp.databar.ai/mcp

https://mcp.databar.ai/mcp

Claude (web and desktop)

Go to Settings > Connectors, click Add custom connector, name it Databar and paste the URL above. Claude sends you to Databar to authorize your workspace, so you never copy an API key. Then ask: "What can the Databar MCP do?"

Claude Code

Add the server from your terminal with your Databar API key:

claude mcp add --transport http databar https://mcp.databar.ai/mcp --header "Authorization: Bearer YOUR_API_KEY"

claude mcp add --transport http databar https://mcp.databar.ai/mcp --header "Authorization: Bearer YOUR_API_KEY"

claude mcp add --transport http databar https://mcp.databar.ai/mcp --header "Authorization: Bearer YOUR_API_KEY"

claude mcp add --transport http databar https://mcp.databar.ai/mcp --header "Authorization: Bearer YOUR_API_KEY"

Cursor and other MCP clients

Add the server to your client's MCP config, for example .cursor/mcp.json:

{
  "mcpServers": {
    "databar": {
      "url": "https://mcp.databar.ai/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}

{
  "mcpServers": {
    "databar": {
      "url": "https://mcp.databar.ai/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}

{
  "mcpServers": {
    "databar": {
      "url": "https://mcp.databar.ai/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}

{
  "mcpServers": {
    "databar": {
      "url": "https://mcp.databar.ai/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}

You'll find your API key under Integrations in your Databar workspace. The hosted server supports Streamable HTTP and legacy SSE, so it works with Codex, Gemini CLI and other MCP-compatible clients too. If your client can only launch a local process, there's also an npm package, databar-mcp-server, but it can lag behind the hosted version. Setup details for every client are in the MCP server docs.

Details that matter once you're running it

• Spending guardrails. By default the server checks your credit balance before each enrichment, and you can set a per-request cost cap, so an agent can't quietly burn through your credits.

• Caching. Enrichment results are cached for 24 hours, and cached lookups don't use credits. If you need a fresh result, ask for one and the cache is skipped.

• Retrieve results promptly. Results from asynchronous jobs are kept for 24 hours after completion, so pull them into a table or your own system before then.

• Agent Skills. The server ships with ready-made skills for single lookups, table-driven enrichment, waterfalls and bulk runs, so your agent knows the full workflow without step-by-step instructions.

MCP, SDK or API: which to use

They all call the same enrichments, so a common pattern is to work a workflow out through the MCP server, then move the version that works into the SDK or API when it needs to run on a schedule without an agent in the loop. We cover the trade-offs in MCP vs SDK vs API and the GTM engineer's guide to the Databar API and MCP server.

What it costs

Connecting the MCP server costs nothing extra. Enrichments use credits from your plan the same way they do in the app or through the API. Plans start with Build at $99 a month for 5,000 credits, and Scale is $495 a month for 50,000 credits; see pricing for what each plan includes. A 14-day trial of the full product comes with 100 credits.

FAQ

What is MCP?

The Model Context Protocol is an open standard for connecting AI assistants to external tools and data. An MCP server tells the assistant which tools exist and what inputs they need, so the assistant can call them directly instead of you writing integration code.

Which AI tools does it work with?

Any MCP-compatible client. Claude (web, desktop and Claude Code), Cursor, Codex and Gemini CLI all work with the hosted server.

Which enrichments can I use through MCP?

The same catalog you see in Databar. If you haven't connected your own key for a bring-your-own-key provider, that provider won't appear in search results until you do.

Do I still need the Python SDK?

Not for conversational or ad-hoc work. For scheduled pipelines and production jobs that run without an agent, the SDK or API gives you more control over retries, concurrency and error handling.

Can I use MCP for large lists?

Yes. Bulk enrichment and bulk waterfall tools process many records in one call, and table tools let you run an enrichment across every row of a table. For very large recurring jobs, the API or SDK is usually the better fit.

Related reading

• Best MCP Servers for Claude Code: 10 That GTM Teams Use

• How Waterfall Enrichment Works: Provider Order, Coverage and Cost

• New Pricing: Built for How GTM Teams Work

Start free with a 14-day trial of the full product, or book a founder demo to see the MCP server run on your own data.