MCP Server

The MCP integration turns your FutureBase knowledge base into a Model Context Protocol server. Any MCP-compatible client — Claude Desktop, Cursor, Windsurf, VS Code with Copilot, and others — can query your knowledge base as a tool, giving AI assistants real-time access to your documentation, FAQs, and support content.

Key Features

• Zero-config server — no CLI, no Docker, no self-hosting. Your MCP endpoint is ready instantly after install
• Streamable HTTP transport — uses the latest MCP spec (Streamable HTTP), compatible with modern clients
• Public or private access — toggle between open access and Bearer-token-protected endpoints
• Rate limiting — built-in 60 requests/minute rate limit per endpoint
• Knowledge base search — exposes a getInformation tool that searches your entire knowledge base

How It Works

1. You install the MCP integration from your dashboard
2. A unique MCP endpoint URL is generated for your organization
3. You add the endpoint URL to any MCP-compatible client
4. When a user asks the AI assistant a question, the client calls the getInformation tool on your MCP server
5. The server searches your knowledge base and returns relevant information
6. The AI assistant uses those results to compose an answer

Quick Setup

{
  "mcpServers": {
    "futurebase": {
      "url": "https://api.futurebase.io/v1/mcp/<your-integration-id>"
    }
  }
}

{
  "mcpServers": {
    "futurebase": {
      "url": "https://api.futurebase.io/v1/mcp/<your-integration-id>",
      "headers": {
        "Authorization": "Bearer <your-token>"
      }
    }
  }
}

Client Setup

Claude Desktop

1. Open Claude Desktop and go to Settings → Developer → Edit Config
2. This opens the claude_desktop_config.json file
3. Add the mcpServers entry shown above
4. Save the file and restart Claude Desktop
5. You should see onedollarchatbot listed under the tools icon (hammer) in the chat input

Cursor

1. Open Cursor and go to Settings → MCP
2. Click Add new MCP server
3. Choose Type: HTTP and paste your endpoint URL
4. If your endpoint is private, add the Authorization header
5. The getInformation tool will be available in Cursor's AI features

Windsurf

1. Open Windsurf and navigate to Settings → MCP Servers
2. Click Add Server and paste the endpoint URL
3. Configure authentication headers if using a private endpoint

VS Code (GitHub Copilot)

1. Open VS Code and go to Settings
2. Search for mcp and locate the MCP Servers configuration
3. Click Edit in settings.json and add:

{
  "mcp": {
    "servers": {
      "futurebase": {
        "type": "http",
        "url": "https://api.futurebase.io/v1/mcp/<your-integration-id>"
      }
    }
  }
}

4. For private endpoints, add the Authorization header in the configuration

Other Clients

Any client that supports the MCP Streamable HTTP transport can connect to your endpoint. Provide the endpoint URL and, for private endpoints, a Bearer token in the Authorization header.

Available Tools

The MCP server exposes a single tool:

getInformation

Searches your knowledge base for relevant information.

Returns: One or more text blocks containing the most relevant content from your knowledge base, including documentation, FAQs, and support articles.

Example usage by an AI assistant:

User: "How do I reset my password?"
→ AI calls getInformation({ question: "How do I reset my password?" })
→ Server returns relevant knowledge base articles
→ AI composes an answer using the retrieved content

Configuration Options

Public vs Private Access

Control who can query your MCP endpoint:

Toggle this setting from the MCP integration settings dialog in your dashboard.

Rate Limits

Each MCP endpoint is rate-limited to 60 requests per minute. If the limit is exceeded, the server responds with a 429 Too Many Requests error and includes a retryAfter value.

If you need higher limits, contact support at support@futurebase.io.

Security

• Stateless architecture — each request creates a fresh server instance; no session data is stored
• Bearer token authentication — private endpoints require a valid token in every request
• Rate limiting — prevents abuse with per-endpoint request throttling
• No credential storage — the MCP integration doesn't require OAuth or store third-party credentials
• Soft-delete uninstall — uninstalling the integration disables the endpoint immediately with a 30-day data retention window

Troubleshooting

Frequently Asked Questions