eToro MCP Server

Connect any MCP-compatible AI assistant (Claude, Cursor, and others) to your eToro account for portfolio analysis, market research, and trading.

Overview • Installation • Configuration • Tools

Overview

MCP server wrapping the eToro public API, exposing 35 tools for AI assistants (Claude Desktop, Cursor, Claude Code, etc.). Built for investment research and portfolio analysis — assess positions, research instruments and themes, vet popular investors for copy-trading, backtest DCA strategies against historical data. Trading is fully supported, but use care: letting an AI agent place real-money orders is genuinely risky, and you should treat it as such.

What Can You Do With It?

The server is built around research workflows first. The examples below show real multi-tool conversations — the assistant chains several etoro_* calls together to answer a single question. Numbers are representative; actual responses draw from live eToro data.

1. Portfolio health check

The assistant calls etoro_get_portfolio to pull your open positions, then etoro_get_instruments to resolve instrument IDs to names and sectors, and etoro_get_rates for live prices. It assembles a full snapshot from the raw data.

You: How is my portfolio doing? Break it down by sector and flag any trouble spots.

Assistant: I pulled your portfolio (98 open positions), resolved each instrument, and fetched live rates. Here's the picture:

• Cash: $22,265 • Invested: $45,782 • Positions: 98 open

Sector breakdown:

Sector	% of invested	Top holdings
Technology	38%	GOOG, META, MSFT, NVDA, AMZN
ETFs & Index	22%	VOO, QQQ, SOXX, ARGT, IEUR
Crypto	15%	BTC, ETH
Healthcare	10%	UNH
Financials	8%	SAP
Other	7%	LDO + others

<img src="https://raw.githubusercontent.com/gabrielcerutti/etoro-mcp-server/main/readme-assets/portfolio-allocation.png" width="480" alt="Portfolio sector allocation" />

Trouble spot: Your tech sleeve is 38% of invested capital and concentrated in 5 mega-caps. A single sector drawdown hits hard — consider whether the ETF overlap (QQQ, SOXX) is intentional diversification or accidental doubling-down.

Tools used: etoro_get_portfolio → etoro_get_instruments → etoro_get_rates

2. Instrument research

Search is keyword-based — etoro_search_instruments finds instruments by ticker or company name, then you chain etoro_get_instruments for full details and etoro_get_candles for price history.

You: Pull up NVDA — I want the current price and a weekly chart for the last month.

Assistant: Searching for "NVDA" and pulling candle data:

• NVIDIA Corporation (NVDA) — InstrumentID 1545
• Live rate: Bid $208.14 / Ask $208.40
• Exchange: NASDAQ • Type: Stock • Sector: Semiconductors

Weekly candles (last 5 weeks):

Week of	Open	High	Low	Close
Mar 22	$172.65	$181.20	$167.01	$167.61
Mar 29	$165.26	$177.47	$162.95	$176.72
Apr 5	$176.12	$189.98	$173.66	$188.53
Apr 12	$184.15	$201.69	$183.98	$201.64
Apr 19	$199.02	$210.92	$197.23	$208.26

<img src="https://raw.githubusercontent.com/gabrielcerutti/etoro-mcp-server/main/readme-assets/nvda-candles.png" width="560" alt="NVDA weekly candlestick chart" />

NVDA rallied +27% off the March low ($162.95 → $208.26), with each weekly candle printing a higher low. Momentum is strong but the move is getting extended.

Tools used: etoro_search_instruments → etoro_get_instruments → etoro_get_rates → etoro_get_candles

3. Copy-trading due diligence

Vet popular investors using real public data. The assistant calls etoro_get_user_profile for identity and stats, then etoro_get_user_performance for the full yearly return history.

You: Vet user jeppekirkbonde for a long-term copy allocation. I want the honest risk picture.

Assistant: Here's what the data says on jeppekirkbonde:

• Name: Jeppe Kirk Bonde • PI Level: 6 (highest tier)
• Copiers: 25,000+ • AUM: $200M+
• Background: MSc Finance, Copenhagen Business School

Year-by-year performance (real data):

Year	Return		Year	Return
2013	+8.9%		2020	+36.9%
2014	−14.3%		2021	+15.9%
2015	+22.4%		2022	−19.2%
2016	+81.1%		2023	+23.9%
2017	+150.9%		2024	+21.0%
2018	−10.9%		2025	+37.7%
2019	+45.1%

<img src="https://raw.githubusercontent.com/gabrielcerutti/etoro-mcp-server/main/readme-assets/jeppe-performance.png" width="600" alt="jeppekirkbonde yearly performance" />

Average annual return: +30.7% across 13 full years. Three down years (2014, 2018, 2022) with a max drawdown of −19.2% in the 2022 tech selloff.

What you should know:

1. This is a concentrated, long-only US tech strategy. The +150.9% in 2017 and +81.1% in 2016 are outliers — don't expect those every year.
2. Down years are real: 2022 cost −19.2%. If you can't stomach a 20% paper loss, this isn't the right copy.
3. 25K+ copiers and $200M+ AUM means his moves have market impact in smaller names. Slippage is a factor.

Reasonable for multi-year capital if you accept the volatility. Not appropriate for money you need inside 18 months.

Tools used: etoro_get_user_profile → etoro_get_user_performance

4. Watchlist management

The full watchlist lifecycle — create, populate, rename, and curate — all through natural language.

You: Create a "Semiconductor" watchlist and add NVDA, AMD, ASML, TSM, and AVGO to it.

Assistant: Done:

1. Created watchlist "Semiconductor" (ID: 48291)
2. Searched and resolved all 5 tickers to instrument IDs
3. Added: NVDA (1545), AMD (2842), ASML (5765), TSM (3901), AVGO (6192)

Your watchlists now:

Watchlist	Items
Recently Invested	38
My Watchlist	34
Stocks pre-screening	69
ETFs	54
Semiconductor ← new	5
People & Smart Portfolios	26
Crypto	6
Commodities	5

Tools used: etoro_create_watchlist → etoro_search_instruments → etoro_add_watchlist_items → etoro_get_watchlists

5. DCA backtest with historical candles

Use etoro_get_candles to pull years of OHLCV data, then simulate a dollar-cost averaging strategy — no orders placed, pure research.

You: Backtest a $500/month DCA into QQQ over the last 3 years. What would I have today?

Assistant: Pulling weekly QQQ candles (156 weeks) and simulating $500 on the first trading day of each month:

• Total invested: $18,000 (36 months × $500)
• Units accumulated: ~38.2 QQQ
• Current value: $25,365 (at $664.01/share)
• Gain: +$7,365 (+40.9% on invested capital)
• Approx. IRR: ~12.8%/year

Worst stretch: Nov 2022 – Mar 2023, paper drawdown of −22% on the accumulated stack. You bought through it — those discounted shares are now your biggest winners.

No orders placed. This is research only.

Tools used: etoro_get_candles → etoro_get_rates (for current price)

6. Macro-aware portfolio review

Combine portfolio data with the assistant's general knowledge to get a risk assessment tied to your actual holdings.

You: My portfolio is heavy on tech. Flag any risks from this week's macro calendar.

Assistant: Your tech exposure totals ~$17,400 (38% of invested capital) across GOOG, META, MSFT, NVDA, AMZN and the SOXX/QQQ ETFs. This week's watch-list:

• Wed — CPI print (core consensus +0.3% MoM). A hot number re-prices rate-cut odds → tech multiples compress first.
• Thu — Fed speakers (Williams, Waller). Hawkish language strengthens USD, pressuring mega-cap international revenue.
• Fri — NVDA earnings (after close). Binary event for your largest single-stock position.

What this means for you:

• A hot CPI + hawkish Fed combo could knock 3–5% off your tech sleeve short-term.
• NVDA earnings will drive ~8% of your portfolio's week-over-week change. Know your thesis before it prints.

Tools used: etoro_get_portfolio → etoro_get_instruments → etoro_get_rates (+ assistant's macro knowledge)

⚠️ Letting an AI execute real-money trades is risky. The same natural-language interface that makes research fast also makes it easy to misinterpret intent — a typo, an ambiguous number, a stale context from earlier in the conversation. Mitigations: start in ETORO_TRADING_MODE=demo, use read-only API keys for anything that doesn't need write access, and require explicit confirmation in your conversation before any real order.

All examples above are research-only. When you're ready to execute, the same interface places orders. Orders always respect the configured ETORO_TRADING_MODE (demo for paper, real for live).

You: Buy $500 of NVDA at market.

Assistant: Confirmed — opening a market position:

• Instrument: NVDA (1545)
• Amount: $500
• Rate: $208.40 (ask)
• Mode: demo

Position opened — ID 3418072. This is a demo trade, no real money involved.

Full trading coverage includes: open position by amount or units (etoro_open_position), close or partially close (etoro_close_position), limit/entry orders (etoro_place_limit_order), cancel pending orders (etoro_cancel_order), and listing pending orders (etoro_get_orders).

Composing with Other MCP Servers

MCP servers are designed to be composed. The eToro server handles portfolio data, live rates, candles, and trading — but an AI assistant can call tools from multiple servers in a single conversation, filling in the gaps that no single API covers.

Example: earnings risk check across two servers

With eToro + a financial data server (e.g. Financial Datasets, Alpha Vantage, or similar):

You: Do any of my open positions have earnings this week? Flag the ones I should watch.

Assistant: Pulling your portfolio from eToro (98 positions), then checking the earnings calendar:

• NVDA — reports Thursday after close. Your position: $4,200 (9.2% of invested). This is a binary event — expect ±8% move.
• GOOGL — reports Tuesday after close. Your position: $3,100 (6.8%). Consensus is +12% YoY revenue; ad revenue mix is the swing factor.
• UNH — reported Monday pre-market (already priced in). Beat on EPS, slight miss on membership growth.

The rest of your holdings have no earnings this week. Your biggest event risk is NVDA Thursday — it's your largest single-stock position.

This isn't possible with the eToro server alone — it knows your positions but not the earnings calendar. The financial data server knows the calendar but not your portfolio. Together, the assistant connects both.

Natural pairings

Configuration

Just add multiple servers to your MCP config — the assistant sees all tools from all servers:

{
  "mcpServers": {
    "etoro": {
      "command": "npx",
      "args": ["-y", "etoro-mcp-server"],
      "env": {
        "ETORO_API_KEY": "your-api-key",
        "ETORO_USER_KEY": "your-user-key",
        "ETORO_TRADING_MODE": "demo"
      }
    },
    "financial-data": {
      "command": "npx",
      "args": ["-y", "financial-datasets-mcp-server"],
      "env": {
        "FINANCIAL_DATASETS_API_KEY": "your-key"
      }
    },
    "news": {
      "command": "npx",
      "args": ["-y", "tavily-mcp-server"],
      "env": {
        "TAVILY_API_KEY": "your-key"
      }
    }
  }
}

The assistant will automatically chain tools across servers when a question requires it — no extra wiring needed.

Installation

Claude Desktop (recommended)

Download the latest .mcpb file from the Releases page and drag it into Claude Desktop:

Extensions → drag etoro-{version}.mcpb into the window

You'll be prompted for your API Key, User Key, and trading mode. That's it.

Via npm (Claude Code, Cursor, Continue, Cody, …)

Add this to your client's MCP config — it'll launch the server from the published npm package on demand, no local install needed:

{
  "mcpServers": {
    "etoro": {
      "command": "npx",
      "args": ["-y", "etoro-mcp-server"],
      "env": {
        "ETORO_API_KEY": "your-api-key",
        "ETORO_USER_KEY": "your-user-key",
        "ETORO_TRADING_MODE": "demo"
      }
    }
  }
}

For Claude Code, the equivalent CLI command:

claude mcp add etoro \
  -e ETORO_API_KEY=your-api-key \
  -e ETORO_USER_KEY=your-user-key \
  -e ETORO_TRADING_MODE=demo \
  -- npx -y etoro-mcp-server

Pin a specific version with etoro-mcp-server@1.1.0 in the args if you want stability over auto-updates.

From source (developers)

If you want to hack on the server locally, clone the repo and build:

npm install
npm run build

Then point your MCP client at the built file (dist/index.js):

claude mcp add etoro-mcp \
  -e ETORO_API_KEY=your-api-key \
  -e ETORO_USER_KEY=your-user-key \
  -e ETORO_TRADING_MODE=demo \
  -- node /path/to/etoro-mcp/dist/index.js

Or in claude_desktop_config.json / equivalent:

{
  "mcpServers": {
    "etoro-mcp": {
      "command": "node",
      "args": ["/path/to/etoro-mcp/dist/index.js"],
      "env": {
        "ETORO_API_KEY": "your-api-key",
        "ETORO_USER_KEY": "your-user-key",
        "ETORO_TRADING_MODE": "demo"
      }
    }
  }
}

MCP Registry

This server is listed on the MCP Registry under the name io.github.gabrielcerutti/etoro-mcp-server. The registry is the public index of MCP servers — think of it as a package registry for AI tools, analogous to what npm is for JavaScript libraries. MCP-compatible clients use it to discover and install servers programmatically without users having to hand-copy configuration.

Each tagged release of this repository is auto-published to the registry by GitHub Actions (via OIDC), so the entry always tracks the latest npm version.

Configuration

Trading mode: demo routes all trading calls through eToro's virtual account. Set to real only when you're ready to trade with real money.

Tools (35 total)

All tools are prefixed with etoro_ for namespace isolation when composed with other MCP servers.

Market Data (8)

Portfolio & Trading (7)

User & Discovery (7)

Watchlists (9)

Social Feeds (4)

Disclaimer

This project is an unofficial, community-built integration. It is not affiliated with, endorsed by, or supported by eToro. Use it at your own risk.

• Not financial advice. Nothing in this server, its documentation, or its example outputs constitutes investment advice, a recommendation, or a solicitation to buy or sell any financial instrument.
• You are responsible for your trades. The server can place real-money orders when configured in real mode. Verify every order before confirming, and understand that AI assistants can misinterpret intent.
• API changes may break things. This server depends on eToro's public API, which can change without notice. Always pin a version in production.
• Demo first. Start with ETORO_TRADING_MODE=demo and only switch to real after you're confident in your setup.

Contributing

Contributions are welcome — bug reports, feature requests, and pull requests.

The DeepWiki page gives an auto-generated overview of the codebase architecture — useful starting point before diving in.

1. Fork the repo and create a branch from main
2. Run npm install && npm run build to verify your changes compile
3. Run npm test to make sure existing tests pass
4. Open a PR with a clear description of what changed and why

License

MIT — use it, fork it, build on it.