Connect Claude AI to TradingView Using MCP: Setup Guide

What if your trading assistant could read a chart, screen dozens of markets, and write the Pine Script for a strategy in one conversation? That’s the promise of connecting Claude to TradingView. The plumbing is the Model Context Protocol (MCP), an open standard Anthropic launched in November 2024 that already powers thousands of public MCP servers.

This guide walks through the real setup, end to end. You’ll learn which community servers actually work, the exact config file to edit, and the one limitation nobody mentions: Claude can design a trade, but it can’t pull the trigger. I’ll show you how to close that gap too.

Key Takeaways

• TradingView has no official MCP server — every option is community-built and unofficial, so vet the code first.
• MCP adoption has grown rapidly, with thousands of community servers now available across public registries.
• Claude + MCP reads, analyzes, and writes strategies — but it cannot execute live broker orders on its own.
• Pair Claude with PickMyTrade to route the alerts it designs straight to your broker or prop firm.

What Is MCP and Why Connect Claude to TradingView?

MCP, the Model Context Protocol, is an open standard from Anthropic, launched November 2024, that lets AI assistants like Claude plug into external tools and data. Connecting Claude to TradingView means Claude can read your charts, run screeners, and write strategy code without you copy-pasting anything.

Why does this matter for traders specifically? Because the analysis layer of trading is exactly what large language models are good at. Adoption backs that up: most hedge funds now use AI for market analysis, and generative AI tools have become part of daily work for the people who run them. The analysis layer is where these models shine.

The protocol itself went from niche to standard fast. OpenAI officially adopted MCP in March 2025, making it a cross-vendor format rather than an Anthropic-only trick. That’s why a TradingView “connector” written this year works with Claude today and likely your next AI client too.

The MCP ecosystem has expanded quickly, with thousands of public servers indexed across registries like Smithery, Glama, and PulseMCP. For traders, that means the TradingView integration is one config block away, not a custom engineering project.

What Do You Need Before You Start?

You need four things, and most traders already have three. A current build of Claude Desktop or Claude Code, Node.js (for npm-based servers) or Python with uv (for Python servers), a TradingView account, and about 20 to 30 minutes.

Here’s the fork in the road most tutorials skip. Claude Desktop and Claude Code use different config files. Desktop reads claude_desktop_config.json. Claude Code reads .mcp.json in your home folder or project. Pick your client first, because it changes every step that follows.

When I first set this up, I wasted an hour editing the wrong file. I was running a desktop-app-driving server (which needs Claude Code) but kept editing the Claude Desktop config. Nothing loaded. So save yourself the headache: confirm your client before you touch any JSON.

One more honest note on prerequisites. Some servers drive the TradingView Desktop app directly over Chrome DevTools Protocol on port 9222. Those need the desktop app installed and launched with a debugging flag. Others just call TradingView’s screener data and need nothing extra. Choose based on whether you want chart control or raw data.

Our finding: Across the most-starred TradingView MCP repos, the single most common setup failure isn’t code — it’s editing the wrong client’s config file. Confirm Desktop vs. Code before anything else.

How Do You Connect Claude to TradingView Using MCP?

You connect Claude to TradingView by adding an MCP server block to your client’s config file, then restarting the app — a five-step process that takes under 30 minutes. No official TradingView server exists, so you’ll install a community one. Here’s the exact sequence.

Step 1 — Install prerequisites. For npm-based servers, install Node.js. For Python servers, install Python and uv. Verify with node -v or uv --version.

Step 2 — Open the config file. In Claude Desktop, go to Settings → Developer → Edit Config. This opens claude_desktop_config.json (Windows: %APPDATA%\Claude\; macOS: ~/Library/Application Support/Claude/).

Step 3 — Add the server block. Drop your chosen TradingView server under the mcpServers key. For an npm-based screener server:

{
  "mcpServers": {
    "tradingview": {
      "command": "npx",
      "args": ["-y", "tradingview-mcp-server"]
    }
  }
}

For a Python server run with uv:

{
  "mcpServers": {
    "tradingview": {
      "command": "uvx",
      "args": ["--from", "tradingview-mcp-server", "tradingview-mcp"]
    }
  }
}

Step 4 — Restart Claude completely. Quit fully, then reopen. Config changes only load on a cold start. Don’t just close the window.

Step 5 — Test it. Ask Claude something concrete: “Screen for stocks with a golden cross and RSI under 40.” If the connector is live, Claude calls the tool and returns real data.

MCP config changes require a full client restart to register, and the tools indicator confirms a successful connection. If nothing shows up, you almost always edited the wrong file or skipped the restart — the two failure modes that account for most stuck setups.

Which TradingView MCP Server Should You Use?

There’s no single winner — the right server depends on whether you want chart control or screener data, and there are several actively maintained options on GitHub. Remember, all of them are unofficial. The most popular repo states plainly that it is “not affiliated with, endorsed by, or connected to TradingView.”

A flashy GitHub star count doesn’t guarantee uptime or safety. Treat MCP servers the way you’d treat any unvetted dependency: read the code, check the last commit date, and test on paper first. A server that drives your desktop app has real reach, so the caution is warranted.

Here’s a practical breakdown of what’s out there:

• Desktop-driving servers control the TradingView Desktop app over Chrome DevTools Protocol — chart analysis, Pine Script editing, drawing, alerts, screenshots. Built for Claude Code.
• Screener API servers tap TradingView’s screener with a deep field set (Piotroski F-Score, Altman Z-Score), golden/death-cross detection, and pre-built strategies. Work with Claude Desktop.
• Data + indicator servers in Python pull real-time prices, dozens of indicators, candlestick patterns, and even Reddit sentiment across crypto and equities.

The leading desktop-driving server exposes dozens of tools for chart and Pine Script control, while a popular screener server covers a wide range of screener fields and several pre-built strategies (see tradesdontlie/tradingview-mcp and fiale-plus/tradingview-mcp-server). Match the tool to your goal: chart work or data work.

What Claude + MCP Still Can’t Do

Claude with MCP can read, analyze, screen, and write Pine Script — but it cannot place a single live broker order on its own. That’s the hard limit, and it’s why a pure Claude-plus-TradingView setup stops short of actual trading. The AI is your analyst and coder, not your execution desk.

Why does this gap exist? MCP servers read TradingView data or drive the desktop app. Neither one connects to your brokerage account or your funded prop-firm account. A few crypto-only repos claim auto-execution through one exchange, but that’s bespoke and doesn’t apply to regulated futures, equities, or prop accounts.

This matters because the stakes are real. Only a small minority of retail day traders are consistently profitable long-term, and most lose money in a given year. A disciplined, rules-based system that removes emotion is one of the few edges that helps — and “rules-based” means automated execution, not manual clicking.

So you’ve got a brilliant analyst who can’t reach the order button. How do you connect the analysis to the account? That’s the next piece.

From Analysis to Live Trades: Claude → TradingView → PickMyTrade

You bridge the execution gap by letting Claude write the strategy and alert, then handing the alert to a routing tool that talks to your broker. PickMyTrade does exactly that — it receives a TradingView webhook with a JSON payload and routes the order to brokers and prop firms with low latency, no coding required.

The full loop looks like this. Claude + MCP reads TradingView and writes a Pine Script strategy plus a webhook alert message in JSON. TradingView fires that alert when conditions hit. PickMyTrade catches the webhook and executes the order on Tradovate, Rithmic, IBKR, or a prop firm like Apex, Topstep, or Tradeify.

PickMyTrade routes TradingView alert webhooks to live broker and prop-firm accounts with low latency, serving thousands of traders at around $50 per month — see PickMyTrade. That turns Claude’s analysis into real fills on a funded account — the step MCP alone can’t reach. The AI designs, TradingView triggers, PickMyTrade executes.

In our own testing, the cleanest division of labor is this: use Claude for what it’s great at (idea generation, indicator math, Pine Script drafting), use TradingView as the signal source, and never let an experimental AI strategy hit a live account before it’s run on paper. The handoff to PickMyTrade is where you add risk controls — max contracts, daily loss limits, and prop-firm rule guards.

Generate your first PickMyTrade webhook free. Let Claude write the Pine Script, paste the JSON alert into TradingView, and connect it to your broker or prop firm in minutes — start free at PickMyTrade.

Frequently Asked Questions

Conclusion

Connecting Claude to TradingView with MCP is genuinely straightforward — pick your client, drop in a config block, restart, and test. The harder truth is what comes after: Claude is a brilliant analyst that can’t reach the order button.

• Setup takes under 30 minutes via claude_desktop_config.json (Desktop) or .mcp.json (Code).
• Every TradingView MCP server is unofficial — vet the code, check commit dates, paper-trade first.
• Claude analyzes and writes; it doesn’t execute. That limitation is widely misunderstood, even as MCP adoption keeps climbing.
• PickMyTrade closes the loop, turning Claude’s strategy into live fills on your broker or prop firm.

Ready to make Claude’s analysis actually trade? Connect your first strategy with PickMyTrade and route TradingView alerts to your funded account today.

---

Disclaimer:
This content is for informational purposes only and does not constitute financial, investment, or trading advice. Trading and investing in financial markets involve risk, and it is possible to lose some or all of your capital. Always perform your own research and consult with a licensed financial advisor before making any trading decisions. The mention of any proprietary trading firms, brokers, does not constitute an endorsement or partnership. Ensure you understand all terms, conditions, and compliance requirements of the firms and platforms you use.

---

Also Checkout: Automate TradingView Indicators with Tradovate Using PickMyTrade