Skip to content

Connecting AI tools via MCP

Prerequisites

  • An active Stackify workspace
  • An MCP-compatible AI tool (Claude Desktop, Cursor, or any client supporting MCP)

Connect an AI tool

Step 1 — Find your MCP server URL

Your workspace's MCP server URL is:

https://app.stackify.se/mcp/{workspace_id}

Find your workspace ID in Settings → General.

Step 2 — Authorise the connection

When an AI tool connects to your MCP server for the first time:

  1. The tool redirects you to Stackify to approve access.
  2. Review the requested permissions (read, write, or both).
  3. Click Authorise.

The AI tool receives an access token scoped to your workspace.

Step 3 — Configure the AI tool

Add Stackify to your claude_desktop_config.json:

{
  "mcpServers": {
    "stackify": {
      "command": "npx",
      "args": ["-y", "@stackify/mcp"],
      "env": {
        "STACKIFY_WORKSPACE_ID": "YOUR_WORKSPACE_ID",
        "STACKIFY_API_KEY": "tvk_live_..."
      }
    }
  }
}
  1. Go to Cursor Settings → MCP.
  2. Click Add server.
  3. Enter your MCP server URL.
  4. Complete the OAuth flow.

Point your MCP client at: 

https://app.stackify.se/mcp/{workspace_id}
Use OAuth 2.1 client credentials or API key authentication.

What the AI can do

Once connected, the AI tool has access to these MCP tools:

Tool Description
list_cards List cards in the workspace
get_card Get a card's details and field schema
list_records Read records from a card (with optional filter)
create_record Add a new record to a card
update_record Modify an existing record
search Full-text search across the workspace

Revoke access

  1. Go to Settings → Connected apps.
  2. Find the AI tool in the list.
  3. Click Revoke access.

The tool's access token is immediately invalidated.