MetaTrader Gave Its AI Assistant Hands. Here's What That Means For Your License Check.

The terminal you know just got a brain transplant

I've been staring at MetaTrader release notes for a long time — mostly because release notes are where MetaQuotes quietly tells you the ground is shifting before anyone in the community notices. This summer's notes are not quiet.

On July 23, 2026, MetaQuotes shipped MT5 build 6060 with native support for the Model Context Protocol (MCP) and what they're openly calling "agentic AI" inside the terminal itself. A week later, build 6090 expanded the AI Assistant's toolkit further: it can now add indicators to a chart and pull a full list of available indicators with their parameters, on its own, without a human clicking anything. That's not a chatbot bolted onto the sidebar anymore. That's an agent with hands on the controls of the same terminal your EA lives in.

I want to be careful here, because I'm not in the business of manufacturing panic. Nobody has demonstrated an MCP-powered exploit against a license check, and this is a legitimate, useful feature for traders who want an assistant that can actually do things instead of just describing them. But if you've spent any real time in this industry, you know the pattern: every time the platform gets more programmable, the threat model for anything running inside it gets a little bit bigger too. An agent that can enumerate indicators and manipulate the chart environment today is an agent that, tomorrow, gets asked to "figure out why this EA won't run" — and agents don't get tired, don't get bored, and don't mind reading your OnInit() function fifty times in a row.

The community's already arguing about AI and trading logic

It's not just the platform vendor pushing agentic AI — the traders are living it in real time too. Browsing the MQL5 blogs today, one post opened with a trader's blunt experience trying to get AI to build a strategy: