If you use an AI assistant like Claude in your daily workflow, MCP provides the bridge to connect it to your trading account. Here is what it is, how it works, and why it matters.

What is MCP?

MCP stands for Model Context Protocol. It is an open standard that gives LLMs (large language models) a structured, secure way to connect to external tools and data sources.

Think of it like a universal plug socket. Before MCP, connecting an AI assistant to an external tool required custom code built specifically for that pair. MCP replaces all of that with a single, shared interface. Any AI that supports MCP can connect to any tool that supports MCP, without bespoke work on either side.

It is an open standard, not tied to any single AI provider. The same protocol that lets an AI assistant read your calendar can also let it check your account balance or place a trade.

MCP acts as a universal bridge. One standard connects an AI assistant to any compatible tool.

Why was MCP built?

AI assistants are useful for conversation. But most of the things people need done involve external tools: files, calendars, databases, platforms.

Without a common standard, connecting an AI to each tool required a custom-built bridge. That was slow, inconsistent, and costly to maintain. MCP solves this by giving developers a single protocol to build to. Once your tool supports MCP, any compatible AI assistant can connect to it.

For traders, that means the same AI assistant you use for research or writing can also connect to your brokerage account, using the same open standard.

How does MCP work?

You do not need to understand the technical detail to use MCP. But the broad shape is useful context. When you enable an MCP connection, three things happen in sequence.

Three steps from zero to connected. You stay in control at each stage.

The AI does not act without a prompt from you. It does not run on a schedule or operate without an explicit instruction.

MCP in trading: what it looks like in practice

Here is a concrete example.

You are reviewing your portfolio in a Claude conversation. You ask: "What is my current exposure on EUR/USD?" Your AI assistant queries your account through the MCP connection and returns the position details — size, entry price, current P&L — directly in the conversation.

You decide to tighten your stop. You type: "Move my stop loss to 1.0820." The AI sends that instruction through the MCP connection. The modification is applied. You get a confirmation. You did not open ThinkTrader, locate the position, or modify the order manually.

Claude

ChelseaAI MCP connected

DEMO

What is my current exposure on EUR/USD?

EUR/USD position

SideLong

Size0.10 lot

Entry1.0785

Current1.0842

P&L+$57.00

Move my stop loss to 1.0820.

Stop loss updated

EUR/USD stop set to 1.0820. Modification confirmed by ThinkTrader.

Illustrative demo data — actual results vary. CFDs are complex instruments and come with a high risk of losing money rapidly due to leverage.

A ChelseaAI session inside Claude. Account data and order management in a single conversation, using demo data for illustration.

What MCP is not

MCP is a connection layer. It is not an autonomous trading system. Three distinctions worth understanding:

✕ Not algo trading	✕ Not trading advice	✓ Funds stay protected
Algo trading runs pre-programmed rules automatically, without you in the conversation. MCP-connected AI assistants act only when you give an explicit instruction.	AI assistants do not generate signals or recommendations. They execute what you instruct and return factual account data.	AI assistants cannot withdraw, deposit, or transfer funds. That capability is never exposed at any permission level.

Why "open protocol" matters

MCP is not a proprietary integration. It is an open standard, which means it is not tied to any single AI provider. Any AI assistant that supports MCP can connect to any MCP-compatible tool. You are not locked in. If you change AI assistant, the integration moves with you.

What this means for you as a trader

You can choose your AI assistant based on what suits your workflow, and switch it later without losing your ThinkTrader connection. Different LLMs handle different tasks differently — research, conversational reasoning, structured analysis — and ChelseaAI works with all of them as long as they support MCP. If you start with Claude today and decide to use a different MCP-compatible assistant next year, the integration moves with you. There is nothing to rebuild on the account side.

Claude currently offers the most straightforward setup experience and is available on free and paid plans. Any other MCP-compatible AI assistant will work the same way.

ChelseaAI and MCP on ThinkTrader

ChelseaAI is ThinkMarkets' MCP server — the connection layer between your ThinkTrader account and your AI assistant. It exposes a defined set of capabilities: market data, account information, order placement, modification, and closure. Permissions are set by you, per account, and can be changed or revoked at any time.

ChelseaAI is available on ThinkTrader only. It is not available on MT4 or MT5.

_{ChelseaAI is an AI trading interface connecting AI assistants to your ThinkTrader account. ThinkMarkets does not endorse, recommend, or support any specific AI system and is independent of all AI platform providers. ChelseaAI does not provide investment advice, recommendations, or signals, and AI-generated outputs should not be relied upon as the sole basis for any trading decision. The client remains responsible for every action taken via the integration and by using ChelseaAI agrees to indemnify and hold harmless ThinkMarkets against any losses, costs, or claims arising from its use.}