Connecting Your AI Client to the Blindata MCP Server
Work In Progress!
This page is currently undergoing improvements. We apologize for any inconvenience and appreciate your patience.
This guide provides a complete walkthrough for connecting any AI client that supports the Model Context Protocol (MCP) to your Blindata tenant. By following these steps, you will enable natural language access to your Blindata data directly from your AI tool.
Connection URL Configuration
The Blindata MCP server is accessible through a standardized endpoint that follows this pattern:
https://{your-blindata-host}/mcp-api/v1/tenant/{tenantUuid}/mcp/
URL Components
{your-blindata-host}
: Your Blindata instance URL (e.g.,https://your-company.blindata.io
){tenantUuid}
: Your specific tenant identifier (a UUID format string)
Finding Your Tenant UUID
You can find your tenant UUID in several ways:
- Blindata Web Interface: Navigate to Settings → Account Information
- API Requests: When authenticating with Blindata APIs, the tenant UUID is included in the request headers
- Administrator: Contact your administrator if you don’t have access to this information
Connection Requirements
- HTTPS Required: The MCP server only accepts secure HTTPS connections
- Authentication: All requests must include proper authentication headers or tokens. The available authentication methods depend on your tenant’s configuration
- OAuth Provider: The supported OAuth providers and flows are determined by your tenant’s configuration. Contact your administrator to learn which authentication options are enabled for your tenant
Connecting from a custom client
When connecting your custom AI client to the Blindata MCP server, you have several authentication methods available depending on your setup and requirements:
Authentication Methods
1. Authorization Bearer Token The simplest method uses a Bearer token provided by Blindata. Include the token in the Authorization header:
Authorization: Bearer <your-token>
2. Blindata Headers For direct API access, use the standard Blindata authentication headers:
x-bd-user
: Your usernamex-bd-apikey
: Your secret credential
3. OAuth Provider Access Token For enterprise setups with configured identity providers, you can use access tokens from your identity provider with the Blindata audience. This method requires:
- Registering your client in the identity provider
- Configuring the Blindata audience in your token request
- Fetching tokens with the proper audience scope
For detailed instructions on client registration and token fetching with your specific identity provider, refer to the Authentication Methods Guide .
Note
OAuth Support and Limitations
Blindata MCP acts as a resource server (similar to the main Blindata application). However, there are important limitations to consider:
- Most major OAuth providers do not support OAuth 2.0 Dynamic Client Registration Protocol (RFC7591)
- OAuth 2.0 Authorization Server Metadata (RFC8414) may be not supported
- OAuth 2.0 Protected Resource Metadata (RFC9728) is not currently supported by Blindata MCP, as this would not be compatible with the main authentication providers in use
Due to these limitations, discovery information and OAuth flows must be manually coded in your custom client implementation. The standard OAuth discovery mechanisms are not available for automatic configuration.
For security reasons, Blindata MCP does not implement workarounds or act as an authorization server. We recommend:
- Working with your identity provider to properly configure OAuth flows
- Using standard OAuth flows supported by your provider
- Following security best practices for token handling and client authentication
- Consulting your identity provider’s documentation for proper client implementation
Connecting from Desktop AI Clients
This section provides instructions for connecting desktop AI clients to the Blindata MCP server. You can use any AI client that supports the Model Context Protocol (MCP) to access your Blindata data through natural language queries.
Before you start, please ensure you have the following prerequisites in place:
- An AI Client with MCP Support: You need an AI client (like Cursor, Claude Desktop, or others) that can be configured to use external tools via the MCP standard.
- Node.js: You must have Node.js version 16 or higher installed on your system. You can verify your version by running
node --version
in your terminal. - Blindata Account Credentials: You need a Blindata account with API access. Have the following credentials ready:
API Key
Username
Tenant UUID
These credentials are essential for authenticating your requests and ensuring the AI client can securely access data from your specific Blindata tenant.
Note
Why use the mcp-remote
package?
Using our mcp-remote
package is highly recommended. It acts as a secure bridge, handling the authentication flow between your AI client and the Blindata MCP server. This is especially useful for identity providers that do not yet fully comply with the latest OAuth 2.1 specifications used by the MCP standard.
For a deeper dive into how Blindata handles authentication, see our Authentication Methods Guide .
Step 1: Install the MCP Remote Package
The mcp-remote
package is a command-line tool that forwards requests from your local AI client to the Blindata MCP server, injecting the required authentication headers.
- Open your terminal or command prompt.
- Install the package globally using
npm
:npm install -g mcp-remote
- Verify the installation was successful by checking its version:
mcp-remote --version
Step 2: Configure Your AI Client
Now you need to configure your specific AI client to connect to the Blindata MCP server. Choose your client from the options below:
Cursor Desktop
Cursor provides an interface for managing MCP server configurations. Follow these steps to enable the Blindata integration:
UI Configuration:
- Open Cursor and navigate to Chat Settings
- In the left sidebar, click on Tools & Integrations
- Click Add Custom MCP to create a new MCP server configuration
- Configure the server with the following settings in the newly created file:
{
"mcpServers": {
"blindata-mcp-server": {
"command": "npx",
"args": [
"mcp-remote",
"https://{your-blindata-host}/mcp-api/v1/tenant/{tenantUuid}/mcp/",
"--header",
"x-bd-apikey:{your-api-key}",
"--header",
"x-bd-user:{your-email}"
]
}
}
}
Claude Desktop
Claude Desktop also offers a direct approach to MCP server management. Here’s how to enable the Blindata integration:
UI Configuration:
- Open Claude Desktop and navigate to Menu > File > Settings
- Look for the Developer section
- Click Edit Config, this will open your File Explorer, open the claude_desktop_config.json configuration file
- Enter the following configuration:
Note
Note: ensure the API key is surrounded by single quotes as shown.
{
"mcpServers": {
"blindata-mcp-server": {
"command": "npx",
"args": [
"mcp-remote",
"https://{your-blindata-host}/mcp-api/v1/tenant/{tenantUuid}/mcp/",
"--header",
"x-bd-apikey:'{your-api-key}'",
"--header",
"x-bd-user:{your-email}"
]
}
}
}
- Exit Claude Desktop (either with Menu > File > Exit or by terminating the activity in your Task Manager) to ensure the config file is reloaded with the new mcp server
Info
Claude Desktop requires a premium subscription to connect to remote MCP servers. The free version only supports local MCP servers running on your machine. Make sure you have an active Claude Pro subscription before attempting to connect to a remote Blindata MCP server.
Other AI Clients
For other AI clients that support the Model Context Protocol, follow these general steps to enable the Blindata integration:
Locate Your AI Client’s MCP Configuration File:
The configuration file is typically stored in your user’s application data directory. The exact path depends on your operating system and the specific AI client.
- Windows:
%APPDATA%
- macOS:
~/Library/Application Support/
- Linux:
~/.config/
Add the Blindata MCP Server Configuration: Once you have located the configuration file or settings panel, add the Blindata MCP server as a new tool. Paste your information according to the Client’s specifications
Using the AI Client’s Settings: Many AI clients provide a user interface to manage tools and integrations.
- Open your AI client and navigate to its Settings or Preferences menu.
- Look for a section related to Tools, Integrations, or MCP Servers.
- You will likely find an option to add or edit the MCP server configuration directly from the interface.
Troubleshooting
If you encounter issues connecting your AI client to the Blindata MCP server, here are some common solutions:
Required Tools
- Ensure Node.js and NPM are installed on your system to use the mcp-remote package
- Verify installation by running
node --version
andnpm --version
in your terminal
Tools Not Appearing
- For Cursor: Try manually saving the mcp.json file again
- For Claude Desktop: Navigate to the Developer section multiple times or exit and relaunch