How to Connect FXMacroData to ChatGPT Custom GPTs

By the end of this guide your Custom GPT will be able to answer live macroeconomic questions — policy-rate history, upcoming central-bank release dates, COT positioning, precious-metals prices, FX spot rates — without you writing a single line of backend code. All it takes is one OpenAPI Actions block pointing at the FXMacroData API.

Path A — OpenAPI Actions Recommended for Custom GPTs

Custom GPTs use Actions to call external APIs. You provide an OpenAPI 3.x schema and ChatGPT maps every operation in that schema to a callable tool. FXMacroData publishes a public OpenAPI schema that describes all its endpoints, so a single import step gives your GPT access to the entire data catalogue.

Step 1 — Open the Custom GPT builder

In ChatGPT, click Explore GPTs in the left sidebar, then + Create in the top-right corner. Switch to the Configure tab to access the schema and actions settings.

Step 2 — Add FXMacroData as an Action

Scroll to the Actions section at the bottom of the Configure tab and click Add actions. On the actions editor page, click Import from URL and paste the FXMacroData OpenAPI schema URL:

https://fxmacrodata.com/api/openapi.json

Click Import. ChatGPT fetches the schema and automatically populates the list of available operations. You should see endpoints for announcements, the release calendar, FX spot rates, COT data, commodities, and more appear in the actions list.

Step 3 — Configure authentication

Directly below the imported schema, find the Authentication section and click the pencil icon to edit it. Set the auth type to API Key, choose Query parameter as the delivery method, and set the parameter name to api_key. Paste your FXMacroData API key in the value field and save.

With this configuration, ChatGPT appends ?api_key=<your-key> to every request it makes to fxmacrodata.com. The key is stored securely by OpenAI and never exposed in conversation output.

Step 4 — Set the privacy policy URL and save

ChatGPT requires a privacy policy URL for any GPT that calls external APIs. Enter https://fxmacrodata.com/privacy in the Privacy policy field, then click Save to finalise your GPT configuration.

Step 5 — Test the connection

Open the GPT preview panel on the right side of the builder and ask a live macro question. A good first test:

“What is the current USD policy rate and when is the next Fed decision?”

ChatGPT calls the announcements endpoint for usd/policy_rate and the calendar endpoint for usd in the background, then synthesises both results into its answer. The underlying REST calls are:

# Latest USD policy rate series
curl "https://fxmacrodata.com/api/v1/announcements/usd/policy_rate?api_key=YOUR_API_KEY"

# Upcoming USD release schedule
curl "https://fxmacrodata.com/api/v1/calendar/usd?api_key=YOUR_API_KEY"

Representative announcements response:

{
  "currency": "USD",
  "indicator": "policy_rate",
  "data": [
    { "date": "2026-03-19", "val": 4.50, "announcement_datetime": 1742414400 },
    { "date": "2026-01-29", "val": 4.50, "announcement_datetime": 1738184400 },
    { "date": "2025-12-18", "val": 4.50, "announcement_datetime": 1734541200 },
    { "date": "2025-11-07", "val": 4.75, "announcement_datetime": 1731002400 }
  ]
}

Your GPT can now state the current rate and announce the next meeting date with second-level timestamp precision — drawn from live data, not training knowledge.

Step 6 — Try real analyst workflows

Once the connection is confirmed, your Custom GPT is ready for more sophisticated multi-step queries. A few examples:

For each of these prompts, ChatGPT selects the right Action(s) from the imported schema — you never need to specify which endpoint to call or construct a URL. Indicator documentation is at /api-data-docs if you want to understand the full field schema or explore less common indicators like non-farm payrolls or EUR inflation.

Path B — MCP via the ChatGPT Desktop App Alternative

OpenAI has added Model Context Protocol (MCP) support to the ChatGPT desktop app (macOS and Windows). If you are using the desktop app rather than the web-based Custom GPT builder, you can connect FXMacroData as an MCP server instead of configuring OpenAPI Actions. MCP works in the standard ChatGPT chat interface and does not require a Custom GPT.

Option A — Hosted MCP endpoint (zero install)

Add the following entry to your ChatGPT desktop app’s MCP configuration (found under Settings → Extensions → Model Context Protocol or by editing ~/Library/Application Support/ChatGPT/mcp_config.json on macOS):

{
  "mcpServers": {
    "fxmacrodata": {
      "type": "streamable-http",
      "url": "https://fxmacrodata.com/mcp?api_key=YOUR_API_KEY"
    }
  }
}

Option B — Local server via uvx

Use this if you prefer a local process or need to pass the API key through an environment variable rather than in the URL. The mcp-server-fxmacrodata PyPI package is downloaded and executed automatically by uvx:

{
  "mcpServers": {
    "fxmacrodata": {
      "command": "uvx",
      "args": ["mcp-server-fxmacrodata"],
      "env": {
        "FXMACRODATA_API_KEY": "YOUR_API_KEY"
      }
    }
  }
}

After saving the config, restart the ChatGPT desktop app. You can then ask FXMacroData questions directly in the ChatGPT conversation window — the MCP tools are resolved in the background. Type “Ping the FXMacroData server” to verify connectivity before running your first data query.

Summary

You have connected FXMacroData to a ChatGPT Custom GPT in three steps: imported the OpenAPI schema from https://fxmacrodata.com/api/openapi.json, configured query-parameter API-key authentication, and verified the connection with a live policy-rate query. Your GPT can now answer macro questions with real data — not training-set approximations.

Ready to extend the integration? Browse the FXMacroData API docs for the full indicator catalogue across all 18 currencies, or explore the MCP server reference for desktop app configuration options. If you do not yet have a key, subscribe to unlock the complete currency set and all indicators.