Connect Serge to your
LLM client

Pick one of two methods. You don't need both.

Open your Claude Desktop config file (Cursor uses the same JSON shape under its MCP settings):

Add Serge under the mcpServers object. Replace the placeholder with your sk_serge_ key:

{
  "mcpServers": {
    "serge": {
      "url": "https://mcp.serge.ai",
      "headers": {
        "Authorization": "Bearer sk_serge_your_key_here"
      }
    }
  }
}

Using read-only OAuth instead? Skip the config file: in Claude, add https://mcp.serge.ai as a connector and sign in. Provisioning still requires an API key with sites:write.

Restart your client so it loads the new server. The Serge tools should appear in the tool picker.

Confirm the connection is bound to the right workspace by asking your client to call whoami:

Ask Serge whoami — which workspace and scopes does this connection have?

It returns the workspace and scopes. Measurement needs traffic:read. AgentSolo provisioning must show exactly sites:write and traffic:read; if not, create a replacement key and rotate safely.

Use plain language. Your client maps the question to the right tool and fills in the period and domain.

If a tool replies site_not_found, a read-only connection should call list_sites. With a provisioning key, call create_site, paste the returned snippet into the tenant site, deploy, then call check_install until it reports receiving.

What you can ask