Skip to content

Use Dash as an MCP server

Dash exposes an external MCP (Model Context Protocol) server endpoint. This lets other AI tools, like Claude, ChatGPT, or Microsoft Copilot Studio, call Dash as a tool provider. They get access to your workspace’s connected tools, skills, and data through Dash’s permission and approval layer.

You might already use Claude Desktop, ChatGPT, or Copilot Studio for certain workflows. Instead of connecting each of those tools separately to HubSpot, Stripe, Google Sheets, and everything else, you connect them to Dash once. Dash handles the tool connections, permissions, and approvals.

The other AI tool sends a request to Dash. Dash runs it with the same access controls and approval gates as if you’d asked Dash directly in Slack or Teams.

Dash’s MCP server uses JSON-RPC over Streamable HTTP:

  • Endpoint: https://api.dashpup.ai/mcp/platform
  • Auth: Authorization: Bearer dak_YOUR_KEY

You need a Dash API key. See Create API keys to mint one. Choose the scope that matches what the external tool needs:

  • Read — query data, search, read-only operations.
  • Write — everything read can do, plus create and update (subject to Dash’s approval flow).
  • Admin — full access, including workspace management.

Run this in your terminal:

Terminal window
claude mcp add --transport http dash https://api.dashpup.ai/mcp/platform \
--header "Authorization: Bearer dak_YOUR_KEY"

Replace dak_YOUR_KEY with your actual API key.

  1. In Claude Desktop, go to Settings > MCP Servers.
  2. Add a new server with these settings:
    • URL: https://api.dashpup.ai/mcp/platform
    • Transport: HTTP
    • Header: Authorization: Bearer dak_YOUR_KEY
  3. Claude Desktop can now call Dash’s tools.
  1. In ChatGPT, open Settings → Connectors (enable developer mode if needed).
  2. Click Add and choose a custom MCP connector.
  3. Set the server URL to https://api.dashpup.ai/mcp/platform.
  4. Add an Authorization: Bearer dak_YOUR_KEY header.
  5. ChatGPT can now use your Dash workspace’s tools.
  1. In Copilot Studio, add a new MCP connector.
  2. Set the server URL to https://api.dashpup.ai/mcp/platform.
  3. Add an Authorization: Bearer dak_YOUR_KEY header.
  4. Copilot can now call Dash’s tools in your flows.

Any tool that supports the Model Context Protocol over Streamable HTTP can connect. Point it at https://api.dashpup.ai/mcp/platform and pass the API key as a Bearer token in the Authorization header.

Once connected, the other AI tool can use any tool Dash has access to, subject to the same controls:

  • Read tools work automatically (if set to auto in your workspace).
  • Write tools go through Dash’s approval flow. The approval card appears in Slack or Teams, not in the other AI tool.
  • Blocked tools (set to off) are unavailable.

The external AI tool cannot bypass your workspace’s permissions. It acts through Dash, not around it.