Skip to content

Client Setup

There are two ways to connect:

  • Hosted: add https://mcp.auseconmcp.com/mcp as a remote MCP server. Nothing to install, no sign-in and no API key. This is the only option for chat apps such as Claude, ChatGPT and Gemini, which connect to remote servers only.
  • Local: your client launches the server itself with uvx ausecon-mcp-server, using the package published on PyPI. This needs uv installed.

Both give the same tools and the same data.

Each app has a setting for adding a custom MCP server or connector. Add one, paste the URL below, and leave authentication off:

https://mcp.auseconmcp.com/mcp

Menu names and plan requirements change often, so the links below go to each app’s own help page.

On the web, desktop or mobile, add a custom connector with the URL above. Once added on any device it is available on all of them. See Getting started with custom connectors.

In Claude, data results are shown as interactive charts.

Add a custom MCP server with the URL above. Custom MCP servers are set up in ChatGPT on the web and then appear in the desktop and mobile apps. Which plans can add one, and whether a workspace admin must approve it, is set by OpenAI. See Connecting custom MCP servers in ChatGPT.

Add a custom app with the URL above in the Gemini web app at gemini.google.com. Once connected it also works in the Gemini mobile app. Availability depends on your Google account, plan and region. See Connect and manage custom apps.

ChatGPT and Gemini receive the same JSON response as any other client. Inline charts have been tested in Claude only.

Terminal window
claude mcp add --transport http ausecon https://mcp.auseconmcp.com/mcp
Terminal window
codex mcp add ausecon --url https://mcp.auseconmcp.com/mcp

This is for OpenAI’s command-line tool. The Codex desktop app is now part of the ChatGPT desktop app; for that, use the ChatGPT steps above.

Terminal window
gemini mcp add --transport http ausecon https://mcp.auseconmcp.com/mcp

Any other client that supports remote MCP servers over Streamable HTTP takes the same URL.

Use these if you would rather run the server on your own machine. The hosted server and a local one return the same results.

Use this shape in Claude Desktop’s MCP configuration:

{
"mcpServers": {
"ausecon": {
"command": "uvx",
"args": ["ausecon-mcp-server"]
}
}
}

The repository includes the same example at examples/claude_desktop_config.json.

Terminal window
claude mcp add --transport stdio ausecon -- uvx ausecon-mcp-server
Terminal window
codex mcp add ausecon -- uvx ausecon-mcp-server

Add an entry to ~/.cursor/mcp.json, or to the project-level .cursor/mcp.json:

{
"mcpServers": {
"ausecon": {
"command": "uvx",
"args": ["ausecon-mcp-server"]
}
}
}

Add an entry to .vscode/mcp.json in your workspace, or the user-level equivalent:

{
"servers": {
"ausecon": {
"type": "stdio",
"command": "uvx",
"args": ["ausecon-mcp-server"]
}
}
}

Smithery deployment uses the checked-in smithery.yaml and Dockerfile.smithery. The container runs ausecon-mcp-http, listens on the PORT environment variable supplied by Smithery, and exposes the MCP endpoint at /mcp.

The HTTP entrypoint is intentionally separate from the stdio command. It is unauthenticated because the server exposes only read-only tools over public ABS, RBA, and APRA data. Do not reuse this pattern for private data sources or write-capable tools without adding authentication and origin controls.

ausecon is an independent open-source project. It is not affiliated with, endorsed by or operated by the Australian Government, the Australian Bureau of Statistics, the Reserve Bank of Australia or the Australian Prudential Regulation Authority. Data is published by those agencies under CC BY 4.0; see Data Sources and Licensing.