How to Connect FXMacroData to Claude / Claude Desktop (MCP)

By the end of this guide Claude Desktop will have live access to macroeconomic indicator series, central-bank release calendars, CFTC COT positioning, precious-metals prices, and FX spot rates across 18 currencies — all queryable in plain English from your Claude Desktop conversations.

Path A — Connect via MCP Recommended

Model Context Protocol (MCP) is Claude Desktop’s native mechanism for connecting to external data sources. One JSON config block and Claude auto-discovers every tool the FXMacroData server exposes — no schema definitions required, no HTTP boilerplate.

Step 1 — Open the Claude Desktop config file

Claude Desktop reads MCP server configuration from a JSON file stored at a fixed path on your machine. Open (or create) the file at the location for your operating system:

On macOS you can open the file directly from Terminal:

open -a "TextEdit" ~/Library/Application\ Support/Claude/claude_desktop_config.json

If the file does not yet exist, create it. If it already exists and contains other MCP server entries, you will add the FXMacroData block alongside them — do not delete existing entries.

Step 2 — Add the FXMacroData MCP server block

Paste the following JSON into claude_desktop_config.json, replacing YOUR_API_KEY with your actual FXMacroData key:

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

The uvx command downloads and runs the mcp-server-fxmacrodata PyPI package automatically each time Claude Desktop starts — no manual pip install needed.

If your config file already has a mcpServers key with other servers, merge the fxmacrodata block into the existing object:

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

Step 3 — Restart Claude Desktop

Fully quit and reopen Claude Desktop so it picks up the new configuration. On macOS:

osascript -e 'quit app "Claude"' && sleep 2 && open -a Claude

On Windows, right-click the Claude icon in the system tray, choose Quit, then relaunch Claude from the Start menu.

After restart, Claude Desktop will launch the uvx mcp-server-fxmacrodata subprocess in the background. If uv is not yet installed on your machine, Claude may show an error; install uv first with pip install uv and then restart Claude.

Step 4 — Verify the connection

Start a new conversation in Claude Desktop and type:

“Use the FXMacroData ping tool to check the connection.”

Claude will call the ping tool. A pong response in Claude’s reply confirms the MCP server is running and reachable. If Claude says the tool is not available, re-check that the JSON is valid (no trailing commas, correct braces) and that Claude Desktop was fully restarted.

Step 5 — Explore the available tools

Once connected, Claude can call the following FXMacroData tools directly:

Claude resolves which tool to call from your natural-language prompt — you never need to name a tool or write JSON. Just ask your macro question and Claude handles the tool dispatch automatically.

Step 6 — Try a real analyst workflow

You are reviewing a potential carry trade on GBP/JPY ahead of the Bank of England’s next decision. In a new Claude Desktop conversation, ask:

“Fetch the GBP and JPY policy rates for the last 12 months, show me the GBP/JPY spot rate over the same window, and summarise the carry differential and which central bank looks more hawkish heading into the next quarter.”

Claude calls indicator_query twice (once per currency) and forex once, then synthesises a carry-trade summary entirely from live data. The underlying REST calls that back these tool invocations are:

# GBP policy rate
curl "https://fxmacrodata.com/api/v1/announcements/gbp/policy_rate?api_key=YOUR_API_KEY"

# JPY policy rate
curl "https://fxmacrodata.com/api/v1/announcements/jpy/policy_rate?api_key=YOUR_API_KEY"

# GBP/JPY spot rate
curl "https://fxmacrodata.com/api/v1/forex/GBP/JPY?api_key=YOUR_API_KEY"

Representative indicator_query response:

{
  "currency": "GBP",
  "indicator": "policy_rate",
  "data": [
    { "date": "2026-03-20", "val": 4.50, "announcement_datetime": 1742482200 },
    { "date": "2026-02-06", "val": 4.50, "announcement_datetime": 1738836000 },
    { "date": "2025-11-07", "val": 4.75, "announcement_datetime": 1731002400 },
    { "date": "2025-08-01", "val": 5.00, "announcement_datetime": 1722506400 }
  ]
}

With the policy rates and spot series in hand, Claude produces a carry analysis grounded in current data rather than training-set knowledge. The GBP policy rate docs and JPY policy rate docs show the full indicator schema if you want to extend the analysis further.

Path B — Direct REST API calls Alternative

If you prefer to feed FXMacroData directly into a Claude system prompt, use it inside a Python script that calls the Claude API, or experiment with data before setting up MCP, the REST API is the simplest path. Every endpoint accepts a query-parameter API key.

Fetching indicator data with Python

import requests

BASE = "https://fxmacrodata.com/api/v1"
KEY  = "YOUR_API_KEY"

# Latest EUR inflation series
resp = requests.get(
    f"{BASE}/announcements/eur/inflation",
    params={"api_key": KEY},
    timeout=10,
)
data = resp.json()
for row in data["data"][:5]:
    print(row["date"], row["val"])

Checking the release calendar

# Upcoming USD releases
resp = requests.get(
    f"{BASE}/calendar/usd",
    params={"api_key": KEY},
    timeout=10,
)
events = resp.json()
for event in events["data"][:3]:
    print(event["indicator"], event["release_date"])

Injecting live data into a Claude API call

You can fetch data from FXMacroData in Python and then pass it to Claude via the Anthropic SDK as part of the user message:

import anthropic
import requests

BASE    = "https://fxmacrodata.com/api/v1"
FX_KEY  = "YOUR_FXMACRODATA_API_KEY"
ANT_KEY = "YOUR_ANTHROPIC_API_KEY"

# Pull the latest USD non-farm payrolls
resp  = requests.get(f"{BASE}/announcements/usd/non_farm_payrolls",
                     params={"api_key": FX_KEY}, timeout=10)
nfp   = resp.json()["data"][:6]

client   = anthropic.Anthropic(api_key=ANT_KEY)
message  = client.messages.create(
    model   = "claude-opus-4-5",
    max_tokens = 1024,
    messages = [{
        "role":    "user",
        "content": f"Here are the last six USD non-farm payroll readings: {nfp}. "
                   "Summarise the trend and tell me what this implies for the Fed outlook.",
    }],
)
print(message.content[0].text)

A full list of supported currencies and indicators is at /api-data-docs. For example, USD non-farm payrolls and EUR inflation pages include the exact endpoint paths, field descriptions, and example responses.

Summary

You have wired FXMacroData into Claude Desktop in three steps: opened claude_desktop_config.json, added the FXMacroData MCP server block with uvx, and confirmed the connection with a ping call. From here every macro question you ask in Claude Desktop triggers a live data call — no browser tab-switching, no manual data imports, no stale training-set figures.

Ready to explore what data is available? Browse the FXMacroData API docs for the full indicator catalogue, or check the MCP server reference for authentication options and the full tool schema. If you do not yet have an API key, subscribe to unlock all 18 currencies and the complete indicator set.