Datafiniti Claude Connector

There are two ways to use Datafiniti with Claude, and both are valid:

1. Connect the MCP server so Claude can run Datafiniti queries directly during a conversation.
2. Just talk to Claude about the data you want — no connection at all — and let Claude help you build a query that you run yourself.

Neither is more "official" than the other. This page covers both, so you can pick based on whether you want Claude to execute queries or just to help you write them. For how this fits alongside other tools, see the Overview.

Option A — Connect the MCP server

When you connect Datafiniti as an MCP server, Claude can list the Datafiniti tools and call them — searching, counting, and starting downloads — inside a normal conversation.

1. Before you start

Connecting from Claude uses OAuth, so you won't paste a token by hand — you'll sign in to Datafiniti through a redirect when you connect. Before you begin, make sure:

• You have a Datafiniti account you can sign in to. Your access and credits are tied to that account exactly as they are for REST.
• You're on a Claude plan that supports remote (custom) connectors. On Team or Enterprise, an organization owner may need to add the connector first, in Organization settings → Connectors; members then connect their own Datafiniti account to it.

2. Add Datafiniti as a connector

In Claude, open Settings → Connectors, click Add (the + button), and choose Add custom connector. In the dialog, enter:

Name:   Datafiniti
URL:    https://api.datafiniti.co/v4/mcp

Leave the Advanced settings (OAuth Client ID / Secret) blank unless Datafiniti has given you specific client credentials to enter there. Click Add to save the connector.

Then click Connect. Claude redirects you to sign in to Datafiniti and authorize access. Once you approve, Claude completes the OAuth handshake over the server's Streamable HTTP transport and can list the Datafiniti tools and prompts.

Finally, start a new chat and enable the connector for that conversation from the + ("Add files, connectors, and more") menu next to the message box.

Screenshot: the "Add custom connector" dialog with the Datafiniti name and URL.

Screenshot: the Datafiniti sign-in and authorize screen reached from Connect.

Screenshot: enabling the connector for a conversation from the + menu.

Team / Enterprise

If your plan is managed, the owner adds the connector once in Organization settings; you'll then find it under your own Connectors, click Connect, and authorize with your Datafiniti account.

Connecting happens from Anthropic's cloud

Claude reaches the MCP server from Anthropic's servers, not your local machine. If Datafiniti's endpoint is ever placed behind an IP allowlist, Anthropic's ranges must be permitted or the connection will fail even though the URL works in your browser.

3. Ask Claude to run a query

With the connector active, ask in plain language:

Find US single-family homes for sale and show me the first few.

Claude will build a Datafiniti query, call df_search, and summarize the records that come back. Ask it to count first (df_count) to size a result set before spending credits, or to start a download (df_start_download) when you want a large pull — Claude polls df_download_status for the result-file links.

Screenshot: Claude calling``_ and summarizing the results._

For the full tool list, defaults, and behavior, see MCP Server.

Credits apply when Claude runs a query

With the connector active, Claude's searches and downloads charge credits exactly as REST does, against the Datafiniti account you authorized. Counting (df_count) is free.

Option B — Just have a conversation

You don't have to connect anything to get real value from Claude here. Because Datafiniti's query syntax is fully documented, Claude can help you construct and refine a correct query in plain conversation — and you run it yourself afterward.

This is a good fit when you'd rather not give a tool live access to your account, when you're still exploring what to ask, or when you just want a second set of eyes on a query.

How to do it

1. Tell Claude what data you're after and which data type (property, people, product, or business).
2. Paste in the relevant context so Claude works from your actual field names — the query-construction guide for your data type, plus the schema or possible-values page for any status or enum fields.
3. Let Claude draft the query, then ask it to explain or refine each part.
4. Run the finished query yourself in Postman, cURL, an SDK, or the Portal.

Example exchange

You: I want US single-family homes currently for sale. Here's the property query guide and the possible status values. Build me a Datafiniti query and explain it.

Claude: country:US AND propertyType:"Single Family Dwelling" AND mostRecentStatus:"For Sale" — this matches US records whose property type is single-family and whose most recent sale-side status is "For Sale." When you send it to the API, escape the quotes: propertyType:\"Single Family Dwelling\".

Screenshot: Claude building and explaining a query in a plain conversation, with no connector attached.

You then run that query in the client of your choice.

Exact-match quoting:" vs ************\"

When a value is quoted for an exact match, which quote form to use depends on where the query goes. In the raw query string, use plain double quotes: FIELD:"value". When that query is a value inside a JSON tool argument or request body, the inner quotes must be escaped as \" so the JSON stays valid: "query": "FIELD:\"value\"". Some coding platforms and clients expect one form over the other — if a query returns an error or no results, check that the quoting matches the context you're pasting into.

Building a query costs nothing

In a plain conversation, Claude only writes the query — no credits are used until you run it yourself.

Which option should I use?

• Choose Option A if you want Claude to run queries and hand you results end-to-end, and you're comfortable authorizing the connector against your Datafiniti account.
• Choose Option B if you want help writing and understanding queries without granting live access, or if you can't connect the server in your environment.

You can also mix them — draft and refine a query in conversation, then connect the server when you're ready to run it at scale.

Worked examples by data type

For step-by-step examples of building a query in conversation and running it — then the connected form where Claude runs it — see:

• Property Examples
• People Examples
• Product Examples
• Business Examples

Next steps

• Full MCP tool reference: MCP Server.
• Test a query by hand: Postman.
• Data-type specifics (fields, statuses, prompts): the "…with Claude" page for Property, People, Product, or Business data.