Docs

Connect CrunchJunkie to VS Code (GitHub Copilot)

Add the CrunchJunkie MCP connector to VS Code with the MCP: Add Server command, sign in with your CrunchJunkie account, and use it from Copilot chat.

What you need

VS Code with its AI features enabled (GitHub Copilot chat) and a CrunchJunkie account on an active paid plan. VS Code's documentation lists no plan requirement for MCP servers, but says your organization can restrict which servers are available. The connector URL is https://app.crunchjunkie.io/api/mcp/mcp. It is also shown under Settings → Connect AI assistants (MCP) in CrunchJunkie.

Add the connector

1. Open the Command Palette and run "MCP: Add Server". 2. Choose "HTTP (HTTP or Server-Sent Events)". 3. Enter the connector URL as the MCP endpoint. 4. Enter a server ID that is not already used, for example crunchjunkie. 5. Choose where to save it. For a personal setup pick "Global", then the follow-up choice VS Code offers (it showed "Copilot Global" when we read the page). To share it with a team, VS Code describes a workspace-root .mcp.json file, which is experimental and sits behind the chat.mcp.workspaceRootConfig.enabled setting. VS Code writes the server into the configuration file you picked. Its documentation points to its MCP configuration reference for the exact file format.

Sign in

When the assistant first uses the connector, it sends you to a CrunchJunkie sign-in page. You sign in with your CrunchJunkie account, the same way you sign in to the app (email and password, Google, or your company's SSO), and approve the access. There is no API key to copy or store. The sign-in is OAuth 2.1 through WorkOS AuthKit, and the assistant registers itself with dynamic client registration or a Client ID Metadata Document, so you enter no client ID or secret. The connection acts as you and sees only what your account can see, in one workspace at a time. Until you choose, that is the workspace you joined first. To change it, open Settings → Connect AI assistants (MCP), or ask the assistant to list your workspaces and switch. If your company requires SSO, connect after signing in with SSO. VS Code's documentation says only to follow the publisher's sign-in instructions if a server needs authentication, so the exact prompt may look different from what is described here.

Check that it works

Start a new chat with the connector switched on and ask: "List my clients." You should get your CrunchJunkie clients back. Then try a real question: "How visible was <brand> in AI answers over the last 30 days, with sample sizes?" Replace <brand> with one of your clients. The answer comes from your live data and quotes how many AI answers each percentage is based on. If the assistant answers from general knowledge instead of calling a CrunchJunkie tool, tell it to use the CrunchJunkie connector. Ask it in a Copilot chat agent session, and check that the tool call shows the CrunchJunkie server as its source.

Write actions

Reading never changes anything. Write actions (creating a report, adding a widget, tracking a prompt and the like) are off until a workspace owner or admin switches on "Allow write actions" under Settings → Connect AI assistants (MCP). Even then, every write tool first returns a preview and changes nothing until you confirm it, and a member with the Read only role cannot make changes at all. The full rules are in the connector documentation.

If it doesn't work

The CrunchJunkie sign-in page never opens, or the assistant says it is not signed in: remove the connector and add it again, and check that the URL is exactly https://app.crunchjunkie.io/api/mcp/mcp. A tool answers "missing scope" or "forbidden": the account you signed in with is not on an active paid plan, or is not a member of the workspace that owns the data. Sign in with an account that has access. If your account belongs to several workspaces, the assistant may be working in another one; ask it to list your workspaces. Write tools are refused: write actions are off by default. A workspace owner or admin switches them on under Settings → Connect AI assistants (MCP). If "MCP: Add Server" is missing, check that AI features are enabled and that you opened a project folder. If your organization restricts MCP servers, an admin has to allow this URL. Other assistants are covered in the connector documentation.