Integrations: Model Context Protocol (MCP) Server

Feedback Falcon is built with engineers in mind. That means we don't want you bouncing between tabs to check on bug reports while you're deep in the code. We provide a robust Model Context Protocol (MCP) server integration to wire your task data directly to AI-enabled IDEs, such as Cursor or WindSurf.

What is an MCP Server?

The Model Context Protocol is a standard for sharing context with AI tools. The Feedback Falcon MCP server exposes an API that allows your agentic IDE tools to interact with your specific Kanban board data.

Setting it up

1. Make sure you have your Feedback Falcon API Key. You can generate one from the MCP Settings Page in your dashboard: https://app.feedbackfalcon.com/settings#mcp.
2. Once generated, you can quickly copy the pre-built configuration JSON tailored for your specific IDE directly from the dashboard.

For Cursor

In Cursor, you can configure your MCP servers by editing ~/.cursor/mcp.json or by using the Settings > MCP Servers panel:

{
  "mcpServers": {
    "feedbackfalcon": {
      "type": "sse",
      "url": "https://app.feedbackfalcon.com/api/mcp/sse?token=YOUR_API_KEY",
      "headers": {
        "X-API-Key": "YOUR_API_KEY"
      },
      "disabled": false,
      "autoApprove": [
        "list_projects",
        "list_tasks",
        "list_boards",
        "get_task_context",
        "inspect_video_segment",
        "create_task",
        "move_task"
      ]
    }
  }
}

For VS Code (Roo Code / Cline)

If you are using VS Code extensions like Roo Code or Cline, the configuration blocks share the same robust format as Cursor. Simply paste the following into your extension's MCP configuration settings:

{
  "mcpServers": {
    "feedbackfalcon": {
      "type": "sse",
      "url": "https://app.feedbackfalcon.com/api/mcp/sse?token=YOUR_API_KEY",
      "headers": {
        "X-API-Key": "YOUR_API_KEY"
      },
      "disabled": false,
      "autoApprove": [
        "list_projects",
        "list_tasks",
        "list_boards",
        "get_task_context",
        "inspect_video_segment",
        "create_task",
        "move_task"
      ]
    }
  }
}

For Claude Desktop (Standard Clients)

For stdio-based clients like Claude Desktop, we recommend using @cloudmcp/connect to proxy the SSE connection to standard I/O:

{
  "mcpServers": {
    "feedbackfalcon": {
      "command": "npx",
      "args": [
        "-y",
        "@cloudmcp/connect",
        "--url",
        "https://app.feedbackfalcon.com/api/mcp/sse?token=YOUR_API_KEY"
      ]
    }
  }
}

Using the Integration

With the MCP Server running, you can open your AI Chat pane in Cursor and ask things like:

"Read the latest bugs assigned to me from Feedback Falcon."

The IDE will connect to the MCP Server and stream your project's feedback tasks directly into your prompt context! It can then view the associated debug logs, read the network errors, and begin suggesting code changes automatically.