Apple Stocks Watchlist MCP Server

Apple Stocks Watchlist MCP Server

An MCP server that lets an AI assistant read your pre-installed macOS Apple Stocks app β€” its watchlist and its cached quotes β€” and open symbols in the app to add them.

Category
Visit Server

README

🍎 apple-stocks-mcp

CI npm version npm downloads MCP Registry platform: macOS Node.js License: MIT

A macOS-only Model Context Protocol (MCP) server that lets an AI assistant read your own data from the Stocks app that ships with macOS β€” your watchlist, and the quotes, fundamentals and intraday charts the app has already cached on your Mac.

100% local and read-only. It reads files that already exist on your own machine, in the Stocks app's container under your home folder. It makes no network requests, uses no external API, needs no API keys, and sends nothing anywhere. The data is whatever the Stocks app last synced β€” open the app to refresh it.

apple-stocks-mcp demo

⚠️ macOS only. The data lives inside the macOS Stocks app's container, so this server does not work on Linux or Windows. On a non-macOS host the server still starts, but exposes a single platform_info tool that politely explains it's macOS-only and how to remove it β€” no failing tools, no crashes.


Quick start (30 seconds)

On a Mac, open the Stocks app once (so its data exists), then:

Claude Desktop / any MCP client β€” add this and restart:

{ "mcpServers": { "apple-stocks": { "command": "npx", "args": ["-y", "apple-stocks-mcp"] } } }

Claude Code:

claude mcp add apple-stocks -- npx -y apple-stocks-mcp

Then ask: "What's the Apple stock doing today?" The first time, your client asks permission to use the tool β€” choose Yes (or "Yes, and don't ask again"). That's it. If anything's off, ask it to run stocks_doctor β€” it tells you exactly what to fix.


Contents


What you can ask

Once connected (see Setup), ask your assistant things like:

  • "What's in my stock watchlist?"
  • "What's the Apple stock doing today?"
  • "Show me the fundamentals for NVDA."
  • "Which of my stocks are up the most today?"
  • "Summarize my whole watchlist."

Privacy

This is the most important section, so it's first.

  • Read-only. The server never modifies, deletes, or writes to the Stocks app's data. It opens the SQLite caches with sqlite3's -readonly flag and only parses the watchlist file. The single write-style action, add_stock, does not touch any file β€” it just asks macOS to open a symbol in the Stocks app so you can add it with a tap.
  • Local only. Every byte read stays on your machine and is only returned to the MCP client you connected (e.g. your local AI assistant). There are no outbound network connections in this server's code β€” you can verify this: there is no fetch/http/https client call anywhere in src/.
  • Your own data. It reads only files owned by your user account, in your home folder (~/Library/Group Containers/group.com.apple.stocks/). It does not access other users' data, remote accounts, or anything outside that container.
  • No telemetry, no analytics, no tracking. None. Ever. The only thing ever written outside the tool results is a one-line startup message to stderr (which platform it started on) β€” that goes to your MCP client's local logs and is never sent anywhere.
  • You stay in control of access. On recent macOS versions the app reading this data (your terminal or MCP client) must be granted Full Disk Access by you, in System Settings. Revoke it any time and the server can no longer read.

Where the data comes from

All paths are inside your own user container ~/Library/Group Containers/group.com.apple.stocks/:

Data File (read-only)
Watchlist symbols Library/Documents/PrivateData/com.apple.stocks.private-production-dbstore.json
Quotes, market cap Library/Caches/shared-database (SQLite, quotes table)
Fundamentals (P/E, EPS, ranges…) same SQLite cache (quote_details table)
Company names / exchanges same SQLite cache (stock_metadata table)
Intraday chart (OHLCV) Library/Caches/sparkline-database (SQLite, sparklines table)

This is the same "read the app's own local store" approach used by other macOS MCP servers such as apple-notes-mcp. It is reading your data from your Mac β€” analogous to exporting your own information β€” and it does not bypass any DRM, access any account, or contact any Apple service.


Tools

Tool What it does
list_watchlist List every ticker in your Stocks watchlist (in order).
get_quote Price + daily change for one or more symbols, from the local cache.
quote_watchlist Cached quotes for every symbol in the watchlist.
stock_details Fundamentals: day & 52-week range, market cap, volume, P/E, EPS, beta, dividend yield, next earnings.
stock_chart Cached intraday OHLCV chart (ASCII sparkline) + change vs previous close.
apple_stock Apple (AAPL) fundamentals + whether it's in the watchlist.
portfolio_summary Whole-watchlist analysis: up/down counts, top gainers/losers, breakdown by currency & exchange.
top_movers The biggest gainers and losers in the watchlist today.
search_watchlist Search the watchlist by symbol or company name.
stocks_doctor Diagnose your setup (macOS? data present? readable? Full Disk Access?).
add_stock Opens a symbol in the Stocks app so you can add it with one tap (no file is written).
platform_info Report whether the server can run here (always available; the only tool on non-macOS).

Resources

Resource URI Contents
stocks://watchlist The watchlist symbols, one per line (text/plain).
stocks://quotes Cached quotes for every watchlist symbol (application/json).

Prompts

Prompt What it does
analyze_portfolio Reviews the whole watchlist and highlights what's notable today.
research_stock Pulls fundamentals + intraday chart for one symbol and summarizes it.

Why add_stock opens the app instead of writing the database

The watchlist lives in an encrypted, iCloud-synced store. Editing it by hand risks corrupting your watchlist and breaking sync, so add_stock deliberately does not write anything β€” it uses the stocks:// URL scheme to open the symbol in the Stocks app, where you add it with one tap and the app keeps its own data consistent.


Example output

These are the actual text results the tools return (symbols shown are generic examples). Numbers come straight from your Stocks app's local cache.

apple_stock / stock_details AAPL β€” fundamentals at a glance:

AAPL (Apple Inc.): 299.18 USD  +3.23 (+1.09%) [open]
  Open: 298.44   Day range: 298.07 – 299.75
  52-week range: 196.86 – 317.4
  Market cap: 4.39T   Volume: 11.12M (avg 47.14M)
  P/E: 36.27   EPS: 8.26   Beta: 1.09   Div yield: 0.36%
  Next earnings: 2026-07-30
AAPL is in the watchlist.

stock_chart NVDA β€” intraday OHLCV as an ASCII sparkline:

NVDA intraday (7 points)
  β–β–β–‚β–ƒβ–†β–‡β–ˆ
  prev close: 204.65   last: 208.69   day range: 207.36 – 208.69
  change vs prev close: +4.04 (+1.97%)

portfolio_summary β€” whole-watchlist analysis:

Watchlist summary β€” 98/99 quoted (1 without a cached quote)
  Up: 57   Down: 41   Flat: 0

  Top gainers:
    NVDA: +1.97%
    AMD:  +4.13%
    MSFT: +0.31%
  Top losers:
    INTC: -2.10%
    ...

  By currency:
    USD: 82
    EUR: 16
  By exchange:
    NASDAQ: 38
    NYSE: 35
    XETRA: 9
    ...

get_quote ["AAPL","MSFT","BTC-USD","EGL.LS"] β€” one line per symbol, across US / crypto / European tickers:

AAPL (Apple Inc.): 299.18 USD  +3.23 (+1.09%) [open]
MSFT (Microsoft Corporation): 378.83 USD  -0.08 (-0.02%) [open]
BTC-USD (Bitcoin USD): 64291.09 USD  -928.50 (-1.42%) [open]
EGL.LS (Mota-Engil, SGPS, S.A.): 4.67 EUR  -0.04 (-0.81%) [open]

What each tool returns

Tool Key fields in the result
get_quote / quote_watchlist symbol, name, price, currency, daily change & %, market state
stock_details the above plus open, day range, 52-week range, market cap, volume & average volume, P/E, EPS, beta, dividend yield, next earnings date
stock_chart per-point open/high/low/close/volume, previous close, day range, change vs previous close, ASCII sparkline
portfolio_summary quoted vs total, up/down/flat counts, top gainers/losers, breakdown by currency and exchange
top_movers ranked gainers and losers (symbol, price, change %)

Requirements

  • macOS, with the Stocks app opened at least once (so its data exists).
  • Node.js β‰₯ 18.
  • The system sqlite3 at /usr/bin/sqlite3 (ships with macOS).
  • On recent macOS versions, the MCP client (or your terminal) may need Full Disk Access to read the Stocks container. If tools return a "not found / unreadable" error, run the stocks_doctor tool, then grant Full Disk Access in System Settings β†’ Privacy & Security β†’ Full Disk Access.

Setup

Option A β€” npx (no install)

Add this to your MCP client config (e.g. Claude Desktop: ~/Library/Application Support/Claude/claude_desktop_config.json):

{
  "mcpServers": {
    "apple-stocks": {
      "command": "npx",
      "args": ["-y", "apple-stocks-mcp"]
    }
  }
}

Option B β€” from source

git clone https://github.com/helderpgoncalves/apple-stocks-mcp.git
cd apple-stocks-mcp
npm install            # also builds via the `prepare` hook
{
  "mcpServers": {
    "apple-stocks": {
      "command": "node",
      "args": ["/absolute/path/to/apple-stocks-mcp/dist/index.js"]
    }
  }
}

Claude Code

Add it as an MCP server:

claude mcp add apple-stocks -- npx -y apple-stocks-mcp

…or install it as a plugin (this repo is also a Claude Code plugin marketplace), which wires up the MCP server for you:

/plugin marketplace add helderpgoncalves/apple-stocks-mcp
/plugin install apple-stocks-mcp@apple-stocks

Restart the client after editing the config.

OpenAI Codex

This repo is also an OpenAI Codex plugin. Add the marketplace and install:

codex plugin marketplace add helderpgoncalves/apple-stocks-mcp
codex plugin install apple-stocks-mcp

Or add it as a plain MCP server in your Codex config (command: npx, args: ["-y", "apple-stocks-mcp"]).


Quick local test

The server speaks JSON-RPC over stdio:

printf '%s\n' \
'{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2024-11-05","capabilities":{},"clientInfo":{"name":"t","version":"1"}}}' \
'{"jsonrpc":"2.0","method":"notifications/initialized"}' \
'{"jsonrpc":"2.0","id":2,"method":"tools/call","params":{"name":"apple_stock","arguments":{}}}' \
| npx -y apple-stocks-mcp

Example output: AAPL (Apple Inc.): 299.18 USD +3.23 (+1.09%) [open]

If anything looks off, run the built-in diagnostics tool β€” ask your assistant to run stocks_doctor, or:

printf '%s\n' \
'{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2024-11-05","capabilities":{},"clientInfo":{"name":"t","version":"1"}}}' \
'{"jsonrpc":"2.0","method":"notifications/initialized"}' \
'{"jsonrpc":"2.0","id":2,"method":"tools/call","params":{"name":"stocks_doctor","arguments":{}}}' \
| npx -y apple-stocks-mcp

FAQ & troubleshooting

The tools' error messages are self-explanatory and tell you exactly what to do, so you rarely need this section. When in doubt, ask your assistant to run stocks_doctor β€” it checks everything and prints the fix.

<details> <summary><b>It asks for permission the first time. Is that normal?</b></summary>

Yes β€” that's Claude's standard safety prompt the first time any MCP tool runs. Choose Yes (or "Yes, and don't ask again" to skip it next time). It's not an error and nothing is sent anywhere. </details>

<details> <summary><b>A tool says the data is "not found" or "unreadable".</b></summary>

The message tells you which case it is. The two causes:

  1. The Stocks app has never run. Open the macOS Stocks app once so it creates its data, then try again.
  2. Full Disk Access. On recent macOS, the app running this server (your terminal, or your MCP client) must be granted Full Disk Access:
    1. System Settings β†’ Privacy & Security β†’ Full Disk Access
    2. Turn it on for your terminal / MCP client (add it with + if missing)
    3. Fully quit and reopen that app, then try again

Run stocks_doctor to re-check. </details>

<details> <summary><b>First call times out or shows "failed to connect".</b></summary>

The very first run downloads the package via npx, which can take a moment. Just try again, or raise the startup timeout: start your client with MCP_TIMEOUT=60000 (milliseconds). </details>

<details> <summary><b>Why aren't the quotes real-time?</b></summary>

This server reads the Stocks app's local cache β€” the last data the app synced. It intentionally makes no network calls. To refresh, open the Stocks app (or its widget); the cache updates and the next tool call reflects it. </details>

<details> <summary><b>A symbol in my watchlist returns "Not in the Apple Stocks cache".</b></summary>

The app hadn't cached a quote for that exact symbol when you asked (some tickers, e.g. certain regional listings, aren't always cached). Open the Stocks app and view the symbol once, then retry. The watchlist still lists it; only its quote was missing. </details>

<details> <summary><b>Can it add a stock to my watchlist automatically?</b></summary>

add_stock opens the symbol in the Stocks app so you add it with one tap. It deliberately does not write to the app's encrypted, iCloud-synced store β€” that would risk corrupting your watchlist and sync. This is a safety decision. </details>

<details> <summary><b>Does it work on Linux or Windows?</b></summary>

No. The data lives in the macOS Stocks app's container, so the server is macOS-only. (The pure parsing logic is unit-tested on Linux CI using synthetic fixtures, but the real data only exists on macOS.) </details>

<details> <summary><b>Is any of my data sent anywhere?</b></summary>

No. There are zero network calls in the source β€” only local sqlite3 (read-only) and open. Everything read is returned solely to the MCP client you connected. See Privacy. </details>


Development & testing

npm run build      # compile to dist/
npm run typecheck  # type-check only
npm test           # build + run the test suite

Tests run against synthetic fixtures in test/fixtures/ (a generated dbstore.json and two small SQLite databases) β€” they never touch your real Stocks data. STOCKS_TEST_MODE=1 bypasses the macOS gate so the pure parsing/formatting logic can be tested on any OS, which is what CI does (GitHub Actions, Linux, Node 18/20/22).

Data paths can be overridden via env vars (used by the tests): STOCKS_DBSTORE_PATH, STOCKS_SHARED_DB_PATH, STOCKS_SPARKLINE_DB_PATH, STOCKS_SQLITE_BIN.

Project layout

src/
  index.ts           # MCP server: tools, resources, prompts
  appleStocks.ts     # reads your Stocks app data (read-only): watchlist + quotes/details/chart
  bplist-parser.d.ts
test/
  appleStocks.test.ts
  fixtures/          # synthetic data β€” no personal holdings
docs/
  demo.tape          # VHS script that generates demo.gif
  run-tool.sh        # helper used by the demo
server.json          # MCP Registry metadata
.github/workflows/ci.yml

To regenerate the demo GIF (requires VHS):

vhs docs/demo.tape   # writes docs/demo.gif

Legal

Not affiliated with Apple

This is an independent, community project. It is not affiliated with, endorsed by, sponsored by, or supported by Apple Inc. "Apple", "Apple Stocks", "Stocks", "macOS", and related marks are trademarks of Apple Inc. They are used here only nominatively β€” to describe, factually and accurately, which app's local data this software reads. No claim of ownership or endorsement is made or implied.

This project ships no Apple code, assets, or trademarks, and does not bundle or redistribute any Apple data.

How it accesses data

The server reads files that already exist on your own Mac, under your own user account, in read-only mode. It does not circumvent any technical protection measure or DRM, does not access any online account or third-party data, and does not transmit data off your machine. The cache file formats may change between macOS versions; if that happens, a tool may simply return "no data" rather than misbehave.

Not financial advice

All figures are read from your Stocks app's local cache, may be delayed or inaccurate, and are provided as-is. Nothing here is investment, financial, legal, or tax advice. Do your own research; do not rely on this software for trading decisions. Market data ultimately originates from third-party providers via the Stocks app and is subject to their terms.

Warranty

This software is provided "AS IS", without warranty of any kind, to the fullest extent permitted by law. See the MIT License. Use at your own risk.


License

MIT Β© HΓ©lder GonΓ§alves

Recommended Servers

playwright-mcp

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.

Official
Featured
TypeScript
Magic Component Platform (MCP)

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.

Official
Featured
Local
TypeScript
Audiense Insights MCP Server

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.

Official
Featured
Local
TypeScript
VeyraX MCP

VeyraX MCP

Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.

Official
Featured
Local
graphlit-mcp-server

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.

Official
Featured
TypeScript
Kagi MCP Server

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.

Official
Featured
Python
E2B

E2B

Using MCP to run code via e2b.

Official
Featured
Neon Database

Neon Database

MCP server for interacting with Neon Management API and databases

Official
Featured
Exa Search

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.

Official
Featured
Qdrant Server

Qdrant Server

This repository is an example of how to create a MCP server for Qdrant, a vector search engine.

Official
Featured