xe-mcp
An MCP server for the Xe Currency Data API that brings live FX rates, historical analysis, and quant-flavored tools into AI tools, working out of the box with zero credentials via Frankfurter/ECB data.
README
xe-mcp
An MCP server for the Xe Currency Data API — brings live FX rates, historical analysis, and quant-flavored tools directly into Claude Code, Claude Desktop, and any MCP-compatible AI tool.
Works out of the box with zero credentials — falls back to Frankfurter (ECB data) automatically. Plug in Xe API keys to switch to Xe's live data. All 12 tools work without credentials.
Tools
| Tool | What it does |
|---|---|
get_rate |
Live mid-market rate between any two currencies |
convert |
Convert an amount at the current rate |
list_currencies |
Common currencies (built-in); full Xe list (~170 currencies) with Xe key |
get_historical_rates |
Daily rates for a currency pair over N days |
volatility_analysis |
Daily std-dev + annualised vol — log-return methodology (FX options standard) |
optimal_send_window |
Percentile rank of today's rate in the N-day distribution + verdict |
nzd_corridors |
NZD snapshot across USD, AUD, EUR, GBP, JPY, SGD, CNY in one call |
correlation_analysis |
Pearson r of daily log-returns between two currency pairs |
rate_alert_check |
Check if a rate has crossed a threshold — returns triggered: YES/NO + distance |
rate_chart |
ASCII line chart of a currency pair's rate history in the terminal |
moving_average |
SMA(20/50/200) with current rate and % distance from each average |
pair_summary |
One-call morning briefing: rate + range + vol + send verdict + SMA(20) |
Live output examples
All examples below use Frankfurter/ECB — no API key required.
> get_rate NZD USD
1 NZD = 0.564940 USD (Frankfurter/ECB, 2026-06-26)
> optimal_send_window NZD USD
NZD→USD send window (Frankfurter/ECB)
Current rate: 0.564940
30-day mean: 0.580598
30-day range: 0.563860 – 0.597400
Percentile: 14th (3/21 historical days were lower)
Verdict: UNFAVOURABLE — bottom quartile
Timestamp: 2026-06-26
> moving_average NZD USD
NZD/USD — moving averages (Frankfurter/ECB)
Current rate: 0.564940
SMA(20): 0.579758 (current is 2.56% below)
SMA(50): insufficient data (need 50 days, have 36)
> volatility_analysis NZD USD 30
NZD/USD — 30-day volatility (Frankfurter/ECB)
Current rate: 0.564940
Range (30d): 0.563860 – 0.597400
Daily volatility: 0.4023%
Annualised vol: 6.39%
Data points used: 21
> correlation_analysis NZD USD AUD USD 30
Correlation: NZD/USD vs AUD/USD (Frankfurter/ECB, 30d)
Pearson r: 0.8553
Interpretation: strong positive
Data points: 20 aligned trading days
> rate_chart NZD USD 30
NZD/USD — last 30 trading days (Frankfurter/ECB)
0.5974 │●●
│ ●●
│ ●●●
│ ●●
0.5806 │ ●●│ │
│ ●●│
│ ●│
│ ●│
0.5639 │ ●●●
└─────────────────────
05-29 06-12 06-26
Setup
Zero-credential mode (Frankfurter/ECB)
All 11 tools work with no API keys using free ECB data via Frankfurter. list_currencies returns a built-in common currency list; with Xe credentials it returns the full ~170 currency list.
git clone https://github.com/CedricConday/xe-mcp
cd xe-mcp
npm install && npm run build
Add to Claude Code:
claude mcp add xe-mcp node /path/to/xe-mcp/dist/index.js
With Xe API (live rates)
Get credentials at xe.com/xecurrencydata, then:
claude mcp add xe-mcp node /path/to/xe-mcp/dist/index.js \
-e XE_ACCOUNT_ID=your_account_id \
-e XE_API_KEY=your_api_key
Or add to .claude/settings.json:
{
"mcpServers": {
"xe-mcp": {
"command": "node",
"args": ["/path/to/xe-mcp/dist/index.js"],
"env": {
"XE_ACCOUNT_ID": "your_account_id",
"XE_API_KEY": "your_api_key"
}
}
}
}
Claude Code slash command
This repo ships a /fx command for quick analysis. Add it to your project:
cp -r .claude/commands /your-project/.claude/
Then use /fx NZDUSD, /fx NZDUSD vol, /fx convert 1000 NZD USD.
Architecture
Local (MCP stdio server)
Claude Code ↔ stdio ↔ xe-mcp ──→ Xe XECD API (if credentialed)
└→ Frankfurter/ECB (free fallback)
xe-mcp/
├── src/
│ ├── index.ts # MCP server (stdio transport)
│ ├── xe-client.ts # Xe XECD API wrapper (authenticated)
│ ├── frankfurter-client.ts # Frankfurter ECB API (free fallback)
│ ├── s3-cache.ts # S3-backed rate history cache (Lambda use)
│ └── tools/
│ ├── rates.ts # get_rate, convert, list_currencies
│ ├── analysis.ts # get_historical_rates, volatility_analysis, optimal_send_window
│ ├── nzd.ts # nzd_corridors
│ ├── correlation.ts # correlation_analysis
│ ├── alerts.ts # rate_alert_check
│ └── chart.ts # rate_chart (ASCII)
├── lambda/
│ ├── handler.ts # REST Lambda — all 10 tools via POST /tool/{name}
│ ├── alert-scheduler.ts # CloudWatch hourly → DynamoDB scan → SQS publish
│ └── alert-processor.ts # SQS consumer → SES email notification
├── src/__tests__/ # 49 unit tests (5 suites)
├── .github/workflows/
│ ├── ci.yml # Test → Build → verify on push
│ └── deploy.yml # Test → Build → SAM deploy to AWS (ap-southeast-2)
├── Dockerfile # Multi-stage Alpine — local & ECS/K8s deployments
└── template.yml # SAM: Lambda + API Gateway + SQS + DynamoDB + S3
AWS deployment (sam deploy)
API Gateway → handler Lambda → Xe/Frankfurter → response
CloudWatch Events (hourly) → alert-scheduler Lambda → DynamoDB → SQS
↓
alert-processor Lambda → SES email
DynamoDB: alert configurations (userId, from, to, threshold, direction)
S3: rate-history cache bucket (scaffolded in src/s3-cache.ts; not yet wired into the handler)
Credential detection: XE_ACCOUNT_ID + XE_API_KEY in env → Xe. Otherwise → Frankfurter. Same fallback in Lambda and locally.
Stack coverage
Built to match the full-stack requirements stated in Xe.com's developer role descriptions. Every item below is backed by code in this repo (S3 caching is scaffolded but not yet wired — noted below).
| Requirement | Where it lives |
|---|---|
| TypeScript | src/, lambda/ — full codebase |
| MCP / agentic tooling | src/index.ts — stdio transport, 12 registered tools |
| AWS Lambda | lambda/handler.ts — REST API over all 10 tools |
| AWS SQS | lambda/alert-scheduler.ts → publishes; lambda/alert-processor.ts → consumes |
| AWS DynamoDB | lambda/alert-scheduler.ts — scans AlertsTable; SAM GSI on userId |
| AWS S3 | src/s3-cache.ts — rate-history cache module + bucket in template.yml (scaffold; not yet wired into the deployed handler) |
| SQLite (local) | src/sqlite-store.ts — rate history cache (RATE_DB_PATH); live in fetchHistoricalSeries — cache hits skip API, misses store to DB; same schema works on PostgreSQL |
| AWS SES | lambda/alert-processor.ts — sends email on alert trigger |
| AWS API Gateway | template.yml — wired to XeMcpFunction |
| CloudWatch Events | template.yml — hourly schedule trigger on AlertSchedulerFunction |
| SAM / IaC | template.yml — full stack as code, sam build && sam deploy |
| CI/CD (GitHub Actions) | .github/workflows/ci.yml — test → build → verify; deploy.yml — deploy to ap-southeast-2 |
| Docker | Dockerfile — multi-stage Alpine build for ECS / local |
| FX domain knowledge | optimal_send_window, volatility_analysis, correlation_analysis — log-return methodology |
Tests
npm test
# Test Suites: 5 passed
# Tests: 49 passed
Tests cover: zero-volatility edge cases, constant-return series, annualised vol formula (× √252), percentile distribution, Pearson r properties (perfect correlation, inverse, zero-variance), NZD/AUD co-movement sanity, SMA computation (period ordering, edge cases, distance from SMA), SQLite schema (PK constraints, upsert behavior, range query ordering, index verification), pair_summary math (percentile ranking, verdict labels, SMA(20) boundary).
Why
Xe's developer job posting requires "daily use of agentic coding tools (Claude Code or equivalent)." I built the tool I'd want if I were working on Xe's FX data pipeline — one that brings rate intelligence into the coding environment without tabbing out.
The quant tools (volatility_analysis, optimal_send_window, correlation_analysis) come from time in currency markets. They're not wrappers around a stock analytics library; the math is direct log-return methodology with test coverage.
Stack
TypeScript · Node.js · @modelcontextprotocol/sdk · Xe XECD API · Frankfurter API
License
MIT
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.