ravn
Guide

Connect Claude to contact data with one URL.

The Model Context Protocol (MCP) is an open standard that lets an AI app like Claude call tools on another service. To give Claude contact data, you add a remote MCP server such as https://api.askravn.com/mcp as a connector and sign in once.

By Ravn · 3 November 2026

Start free: 1,000 lookups Read the docs

The idea

What is MCP, in plain terms?

On its own, a model can only work with what is in the conversation. MCP is a common way for an AI app to discover and call tools that live somewhere else. A server publishes a list of tools, each with a name, a description and a schema for its inputs. The app shows that list to the model. When the model decides a tool would help, the app calls it and passes the result back.

The point of a shared protocol is that one server works with many apps. The same MCP server can be used from Claude, Claude Code, Cursor and other clients without separate integrations for each.

Two kinds

What is the difference between remote and local MCP servers?

A local server is a program that runs on your own computer. The app starts it and talks to it over standard input and output. Local servers suit tools that need your files or your machine, and you install and update them yourself.

A remote server runs on the internet and is reached over HTTP. You add it by URL, and the provider keeps it running and up to date. Remote servers suit services that already live online, such as data APIs, because nothing has to be installed.

Ravn's MCP server is remote, at https://api.askravn.com/mcp, using the Streamable HTTP transport. The MCP server page lists its tools.

Signing in

How does OAuth sign-in work for a connector?

A remote server needs to know who is calling. There are two common ways. The first is an API key sent in a header, which works well in config files and scripts. The second is OAuth, which is what you see when a connector opens a sign-in page and asks you to approve access.

With OAuth, the app never sees a password. It sends you to the provider's sign-in page, you prove who you are and click Allow, and the provider hands the app a token scoped to your account. The flow uses PKCE, a one-time secret that stops anyone who intercepts the redirect from using it. You can revoke the token later without changing anything else.

Ravn supports both. Sign-in is an emailed code followed by an Allow screen, with no password. For config files, a key in the Authorization: Bearer rw_live_… header works too. Keys and tokens are stored hashed.

Set up

How do you add Ravn to Claude?

  1. / 01

    Open connectors

    In Claude, go to Settings, then Connectors, then Add custom connector.

  2. / 02

    Paste the URL

    Use https://api.askravn.com/mcp.

  3. / 03

    Sign in

    Enter your email, type the 6-digit code you receive, and click Allow. A person has to receive the code; an agent cannot sign itself up.

  4. / 04

    Check it worked

    Ask Claude "What Ravn tools do you have?" It should list the tools, including the free account tools such as get_usage.

  5. / 05

    Claude Code

    Run claude mcp add --transport http ravn https://api.askravn.com/mcp, then type /mcp and choose Authenticate.

Config file

What if my client uses a JSON config instead?

Cursor and similar clients read a config file. Use a key in the header.

// .cursor/mcp.json
{
  "mcpServers": {
    "ravn": {
      "url": "https://api.askravn.com/mcp",
      "headers": { "Authorization": "Bearer rw_live_…" }
    }
  }
}
Using it

What can you ask Claude once it is connected?

Ask in plain language. Claude picks the tool and fills in the inputs. Some prompts that work well:

  • "Check whether jane@acme.com is a real mailbox, and tell me the risk grade and why."
  • "Find the work email for Jane Doe at acme.com. Tell me the pattern and how confident you are."
  • "Scrape stripe.com and give me the company description, social links and the technologies it uses."
  • "Verify every address in the attached CSV and split them into send, review and drop."
  • "How many lookups have I used this month?"

Be specific about what you want back. Asking for the reason along with the verdict means Claude reports catch-all and unknown results honestly instead of rounding them up.

Costs and limits

What does a connected Claude cost to run?

Each tool call uses your Ravn plan: every tool is one lookup, and a miss is free. Our own checks such as verification and email finding need no provider key; people and company lookups run on your own provider key (People Data Labs), which bills you for the records. The free plan gives 1,000 lookups to start, with no card, and the Ravn plan is $24 a month with 10,000 lookups included, then $1 per 1,000. See pricing and fair use. If you would rather work from a terminal, the same tools are available through the CLI and REST, documented in the docs.

FAQ

Common questions.

All questions →

Do I need to install anything to use a remote MCP server?

No. A remote server is added by URL. You only install software for local MCP servers, which run on your own computer.

Can I use an API key instead of OAuth with Ravn's MCP server?

Yes. Send the header Authorization: Bearer rw_live_… from any client that supports custom headers, such as a Cursor config file.

Does Claude see my Ravn password?

There is no password. You sign in with an emailed code and approve access, and Claude receives a token tied to your account.

Which tools does the Ravn connector expose?

All 25 tools, from email verification and finding to people and company lookups on your own provider key, plus the free account tools. The MCP server page lists them.

More in blog

Every page uses the same account, plan and tools. See pricing →