Open API / MCP

Yes — when trading is enabled, an AI agent connected through cTrader MCP can place, modify and close real orders on your behalf. This is powerful and risky. Enable the confirmation requirement for trading operations in MCP server settings, supervise the agent and verify outputs before approving sensitive actions. AI-generated trading is not financial advice.

Yes — through cTrader's Local or Remote MCP server. With Claude Code (or Claude Desktop with MCP support) configured to the cTrader MCP server, Claude can read account data, query market data, manage orders and positions, control charts (Local MCP) and execute trading actions if trading is enabled. Trading actions can be gated by a confirmation setting.

Yes. ChatGPT Codex and Cursor both support MCP and can connect to cTrader's Local or Remote MCP server using the same configuration approach as any other supported AI client. Once configured, the AI agent has access to whichever MCP server you point it to.

Yes. Open API supports the operations needed to power a Telegram (or Discord, Slack) trading bot — account checks, market data, alerts, order placement and copy notifications. Start with read-only and notification features before adding trade execution, require confirmations for live orders and log every action so the bot's behaviour is auditable.

Yes — through a webhook bridge. TradingView sends an alert as a webhook to your server, which validates the message, applies risk checks and submits the order to cTrader Open API. TradingView alerts cannot call Open API directly because Open API uses TCP/Protobuf or WebSocket with OAuth-style authentication, not plain HTTP webhooks. Always build the bridge on a demo account first.

Yes. In Local MCP server settings, tick "Require confirmation for trading operations" to gate every trade behind manual approval. Many AI clients also support tool-call confirmation natively — Claude Code, Cursor and others prompt the user before executing tool calls. Use both layers for live trading.

Yes. Open API supports the trading operations available to standard cTrader accounts — placing, modifying and closing orders, plus position and account management. With trading permission, an Open API app can act on a live account exactly as the user could, so authentication, scope and risk controls should be designed before going live.

Yes. Open API works with both demo and live cTrader accounts. Demo is the default starting point for any new integration because real-fund mistakes are unrecoverable. Use a clear separation of credentials and environments between demo and live to avoid mis-routed orders.

cTrader MCP exposes trading, market data, account data, charts, indicators, plugins and workspaces — but cBot backtesting is run inside cTrader Algo on desktop, not as a documented MCP tool. An AI agent connected via Local MCP can open cTrader Algo workspaces and view backtesting results, but driving the backtesting through MCP is not a standard exposed operation.

Yes — Local MCP includes plugin and workspace control as part of its scope, so an AI agent can open, configure and arrange plugins inside cTrader Windows or Mac. Remote MCP does not expose plugin control. For plugin-specific automation, use Local MCP.

Yes — Local MCP can add, tune, remove and read values from built-in indicators on the active chart in cTrader Windows or Mac. This makes it possible for an AI agent to ask "what is RSI(14) on EURUSD H1 right now?" and get a live answer. Remote MCP does not include indicator control.

Yes. C# is a first-class language for cTrader Open API and is also the primary language for in-platform cBots and indicators in cTrader Algo (formerly cTrader Automate). C# developers can choose between building an external Open API application and building a cBot, depending on whether the workflow needs in-platform UI and chart access (cBot) or runs outside cTrader as a service (Open API).

Yes. Python is one of the most common languages used with cTrader Open API, with community libraries and official examples. Typical Python use cases are market data pipelines, research notebooks, dashboards, signal bots and Telegram-style notification bots. For full in-platform automation, a cBot in Python or C# is often a simpler route than Open API.

Start on a demo account, enable confirmation for trading operations and run the agent on read-only tasks first (account checks, market analysis) before allowing it to place orders. Add a hard position-size limit, a max-positions guard and a kill switch in cTrader itself, not just in the AI client. Log every tool call. Move to live trading only after the agent has handled demo workflows without surprises for a meaningful period.

cBots run inside cTrader as in-platform algorithms (cloud or local). Open API runs outside cTrader as a network-connected developer interface (JSON/Protobuf, 50 req/sec on real-time data, 5 req/sec on historical). MCP exposes cTrader capabilities as tools for AI agents to call from natural-language prompts.

Local MCP authenticates through your active cTrader desktop session — if you are signed in to cTrader Windows or Mac, Local MCP inherits that authentication. Remote MCP uses per-account tokens generated in cTrader Web Settings → Remote MCP; each trading account has its own token. Switching accounts requires re-authenticating and updating the token in the AI client.

No. cTrader AI Agent Connect is a tool that lets AI agents interact with cTrader; AI-generated suggestions and actions are not financial, investment, legal or tax advice. The user is responsible for verifying outputs, supervising strategies, securing credentials and using MCP at their own risk.

cTrader Open API enforces operational limits per connection: a maximum of 50 requests per second on real-time and non-historical data and 5 requests per second on historical data. Production integrations should batch requests where possible, avoid sending order modifications on every tick and back off on rate-limit responses. Design rate-aware code from the start; retro-fitting it after a breach is harder.

Local MCP covers a wider scope than Remote MCP because it runs against cTrader Windows or Mac. It adds chart control, UI layout, indicators, plugins, workspaces and price alerts on top of the trading, account and market-data operations available through Remote. Remote MCP is the right choice for cloud or cross-device AI workflows; Local MCP is the right choice for full desktop control.

CLI MCP refers to MCP usage from command-line AI clients like Claude Code, ChatGPT Codex (CLI), Gemini CLI and Cursor. The MCP server (Local or Remote) is the same; the CLI is just one of several client types supported. Configuration is by copying a snippet from cTrader settings into the AI client's config file.

cTrader AI Agent Connect is the official toolkit for connecting AI agents to cTrader through the Model Context Protocol (MCP). It ships with two MCP servers (Local and Remote) plus a skills library that gives agents ready-made workflow instructions. Supported clients include Claude Code, ChatGPT Codex, Cursor and Gemini CLI. AI-assisted trading is not financial advice.

cTrader Open API is a developer interface for building external applications that connect to the cTrader backend — market data, account information, order management, dashboards, trading bots and integrations. It is language-flexible because it speaks JSON or Protocol Buffers and is commonly used with Python or C#. Open API sits alongside cBots (in-platform) and MCP (AI-agent) as one of three developer routes into cTrader.

Local MCP is the cTrader MCP server that connects an AI agent directly to cTrader Windows or Mac, exposing trading, account data, market data, chart control, indicators, plugins, workspaces, UI layout and price alerts as tools the agent can call. It authenticates through your active cTrader desktop session. Setup is via configuration snippet pasted into the AI client.

MCP (Model Context Protocol) is an open specification that lets AI agents discover and use external tools through a standard interface. cTrader exposes its capabilities — trading, account data, market data, charts, indicators, plugins, workspaces — as MCP tools so AI agents can call them from natural-language prompts. Two MCP servers exist: Local (Windows/Mac, widest scope) and Remote (Web, trading and market data).

Remote MCP is the cTrader MCP server that runs through cTrader Web, exposing trading, account data, order and position management and market data analysis as MCP tools. It authenticates through per-account tokens generated in cTrader Web Settings → Remote MCP. Remote MCP works from any device with a supported AI client, without needing the desktop app open.

JSON and Protobuf are two message-serialisation options for Open API. JSON is human-readable and easy to inspect, which helps during prototyping. Protobuf is binary, more compact and faster to parse, which suits production integrations and higher message rates. Pick JSON for early development and Protobuf when latency or throughput matters.

cTrader Open API is language-agnostic because it communicates over JSON or Protocol Buffers. Any language with a TCP/TLS client and Protobuf or JSON support works, including Python, C#, JavaScript/TypeScript, Java, Go and C++. Official SDKs and code samples target Python and C# first.

Useful builds on cTrader Open API include custom trading dashboards, multi-account portfolio views, Telegram/Discord notification bots, signal/research apps, social-trading tools and integrations into external analytics. WebView plugins are the right choice when the goal is a tool embedded inside cTrader; Open API is the right choice when the tool runs outside.

cTrader AI Agent Connect supports any MCP-compatible AI client. Officially documented clients include Claude Code, ChatGPT Codex, Cursor and Gemini CLI. Other clients that implement the MCP specification can also connect. Setup is the same across clients: copy the MCP configuration snippet from cTrader settings into the AI client's configuration.

Anyone with a cTrader ID can register an Open API application and connect through a cTrader-affiliated broker. Open API is aimed at developers building external apps; non-developers are usually better served by cBots, plugins or Copy. Live API access uses real funds, so first integrations should run on demo accounts.