> ## Documentation Index
> Fetch the complete documentation index at: https://docs.fingerprint.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Fingerprint MCP server

> Connect AI assistants and agents to Fingerprint device intelligence

The Fingerprint MCP server connects AI assistants and agents to Fingerprint's device intelligence platform. Once connected, your AI tools can query identification events, detect anomalies, investigate fraud patterns, and manage your Fingerprint workspace through natural language.

The server is available in two forms:

* **Managed service**: Hosted by Fingerprint. No setup or maintenance required.
* **Open source**: Self-host anywhere for full control over deployment and configuration.

This guide covers the managed service. For self-hosting instructions, see the [GitHub repository](https://github.com/fingerprintjs/fingerprint-mcp-server).

## Prerequisites

To use the Fingerprint MCP server, you need an active Fingerprint account and permission to create API keys in your workspace. No additional setup is required before connecting. API keys and workspace access are configured automatically during the OAuth connection flow.

## Connection Details

### Server URL

```
https://mcp.fpjs.io/mcp
```

## Authentication

The Fingerprint MCP server uses OAuth. When you connect your AI client, you will be redirected to the Fingerprint dashboard to authorize access and select the workspace to connect to. If you have only one workspace, the selection step is skipped automatically.

If your AI client does not support OAuth, you can authenticate by passing your API key directly in the client's configuration. Go to the [Fingerprint dashboard](https://dashboard.fingerprint.com/mcp) for configuration options.

## Installation

<Accordion title="Claude (Claude.ai / Claude Desktop)">
  See the [Claude Connector documentation](https://support.claude.com/en/articles/11176164-use-connectors-to-extend-claude-s-capabilities) for general setup instructions.

  <Steps>
    <Step title="Open Connectors">
      Go to **Customize** > **Connectors** and add a custom connector.
    </Step>

    <Step title="Add the server">
      Set **Name** to `Fingerprint` and **URL** to `https://mcp.fpjs.io/mcp`, then click **Add**.
    </Step>

    <Step title="Connect">
      Find Fingerprint in the **Not connected** list and click **Connect**.
    </Step>

    <Step title="Authorize">
      Complete the OAuth authorization in your browser when prompted.
    </Step>
  </Steps>
</Accordion>

<Accordion title="Claude Code">
  See the [Claude Code MCP documentation](https://code.claude.com/docs/en/mcp#installing-mcp-servers) for general setup instructions.

  <Steps>
    <Step title="Add the server">
      Run the following command to add the Fingerprint MCP server:

      ```bash theme={"theme":"github-dark-dimmed"}
      claude mcp add --transport http fingerprint https://mcp.fpjs.io/mcp
      ```
    </Step>

    <Step title="Open the MCP menu">
      Start Claude Code and open the MCP menu:

      ```bash theme={"theme":"github-dark-dimmed"}
      claude
      /mcp
      ```
    </Step>

    <Step title="Authorize">
      Select Fingerprint and complete the OAuth authorization in your browser when prompted.
    </Step>
  </Steps>
</Accordion>

<Accordion title="Cursor">
  See the [Cursor MCP documentation](https://cursor.com/docs/mcp#installing-mcp-servers) for general setup instructions.

  <Steps>
    <Step title="Open Cursor Settings">
      Go to **Cursor** > **Settings…** > **Cursor Settings**.
    </Step>

    <Step title="Add a custom MCP server">
      Navigate to **Tools & MCP** and click **Add Custom MCP**. Add this configuration to your `mcp.json`:

      ```json theme={"theme":"github-dark-dimmed"}
      {
        "mcpServers": {
          "Fingerprint": {
            "url": "https://mcp.fpjs.io/mcp"
          }
        }
      }
      ```
    </Step>

    <Step title="Connect">
      Return to **Tools & MCP** and click **Connect** to authenticate with Fingerprint in your browser.
    </Step>
  </Steps>
</Accordion>

<Accordion title="VS Code (GitHub Copilot)">
  See the [VS Code MCP documentation](https://code.visualstudio.com/docs/copilot/chat/mcp-servers) for general setup instructions.

  **Quick setup:** Open the Command Palette (`Ctrl+Shift+P` / `⇧⌘P`), run **MCP: Add Server**, and follow the guided flow for remote servers.

  **Manual setup:**

  <Steps>
    <Step title="Open your MCP config">
      Create or open `.vscode/mcp.json` in your project (to share with your team), or open the Command Palette and run **MCP: Open User Configuration** to configure across all workspaces.
    </Step>

    <Step title="Configure the server">
      Add the following to your `mcp.json`:

      ```json theme={"theme":"github-dark-dimmed"}
      {
        "servers": {
          "Fingerprint": {
            "type": "http",
            "url": "https://mcp.fpjs.io/mcp"
          }
        }
      }
      ```
    </Step>

    <Step title="Connect">
      Save the file. Click **Start** above the Fingerprint server entry, then complete the OAuth authorization in your browser.
    </Step>
  </Steps>
</Accordion>

<Accordion title="Other MCP Clients">
  Any MCP-compatible client can connect to the Fingerprint MCP server.

  <Steps>
    <Step title="Add the server">
      Follow your client's documentation for adding an MCP server.
    </Step>

    <Step title="Configure the URL">
      Set the server URL to `https://mcp.fpjs.io/mcp`.
    </Step>

    <Step title="Authorize">
      Authenticate with OAuth in your browser when prompted.
    </Step>
  </Steps>
</Accordion>

## Available Tools

Tools are divided into two groups: event tools for querying identification data, and management tools for administering your Fingerprint workspace.

* **Event tools** let you retrieve and search identification events, including Smart Signal data like bot detection, VPN, proxy, incognito mode, and more.
* **Management tools** let you create, update, and delete workspace environments and API keys.

For a full list of tools, see the tool reference in the [GitHub repository](https://github.com/fingerprintjs/fingerprint-mcp-server).

## Example Queries and Prompts

Once connected, you can ask your AI assistant questions in plain language. Some examples:

* Fetch the last 50 events for visitor ID `Abck1234CUFmcnjLwIs4A9` and summarize their activity.
* Find all events linked to order ID `3936532456` and tell me if anything looks suspicious.
* Show me recent events where a bot was detected, filtered to the last 6 hours.
* List all environments in my workspace and create a new one called "staging".

The server also includes an `onboarding` prompt, a guided walkthrough for integrating Fingerprint into a new project that covers JavaScript agent installation and verification steps. Trigger it by asking your AI assistant to walk you through Fingerprint setup.

## Security Considerations

### **Prompt injection**

Fields like `linkedId` and `tag` are customer-controlled values attached to identification events. If a malicious actor controls what gets stored in these fields, they could attempt to embed instructions that manipulate your AI assistant. Most AI clients show you each tool call before it executes, so keep manual approval enabled and review them before confirming.

### **Sensitive operations**

Management tools can create, edit, and delete API keys and environments. Avoid connecting the server in shared or untrusted AI environments. Before approving any destructive changes, particularly in production, take time to carefully review what the agent is proposing.

## Troubleshooting

### Authentication failures

* Ensure you completed the OAuth flow and selected the correct Fingerprint workspace.
* Try disconnecting and reconnecting the MCP server in your client settings.
* If workspace authorization fails, your account may have reached the maximum number of allowed API keys (typically 5). Remove an unused key from the [Fingerprint Dashboard](https://dashboard.fingerprint.com/api-keys) and try again.

### Tools not appearing

* Restart your MCP client after adding the server configuration.
* Try disconnecting and reconnecting the MCP server in your client settings.

### Rate limit errors

* Reduce the number of concurrent or rapid sequential tool calls.
* Use specific filters (`visitor_id`, `linked_id`, time ranges) to reduce the number of API requests needed to answer a question.