Start

Connect an agent

Add the ScopeMCP remote MCP endpoint to Claude, ChatGPT, Codex or Claude Code.

The endpoint

Everything is served from one Streamable HTTP endpoint: https://scopemcp.galavic.world/mcp. Discovery metadata lives at /.well-known/mcp.json, so a client that supports discovery can find the endpoint from the domain alone.

remote MCP endpoint
https://scopemcp.galavic.world/mcp

Before you start

Two things need to be true.

  • You have signed in to ScopeMCP and connected at least one data source.
  • You are using a client that supports remote MCP over Streamable HTTP.

Claude

Add it as a custom connector.

  • Open Settings → Connectors → Add custom connector.
  • Set the URL to https://scopemcp.galavic.world/mcp.
  • Complete the ScopeMCP authorisation in the browser that opens.
  • Ask "what can ScopeMCP do?" to confirm the catalogue loaded.

ChatGPT

Install ScopeMCP as a custom plugin.

  • Open Settings → Apps & connectors → Create a custom app.
  • Add https://scopemcp.galavic.world/mcp as the MCP server URL.
  • Authorise the connector when prompted.
  • Test with "list my Search Console properties".

Codex

Add it to your session configuration.

toml
[mcp_servers.scopemcp]
url = "https://scopemcp.galavic.world/mcp"

Claude Code

Register it alongside your repo tools.

.mcp.json
{
  "mcpServers": {
    "scopemcp": {
      "type": "http",
      "url": "https://scopemcp.galavic.world/mcp"
    }
  }
}

Two ways to authenticate

A browser session works through the cookie. For unattended or CI use, create an API key in Settings → API keys and send it as `Authorization: Bearer <key>`.

Verifying it works

Four prompts that exercise the whole stack in order.

text
1. "What can ScopeMCP do?"          → catalogue loads, no auth needed
2. "List my properties."              → confirms the connection
3. "Health score for the last 28 days." → confirms a data pull
4. "Diff /pricing against the 14 days before." → confirms analysis