What it does
MCP (Model Context Protocol) is an open standard that lets AI clients use tools from other services. Once you connect QuantPilot, your AI client can:
read the QuantScript v2 documentation
look up supported markets and symbols, and the dates they have data for
validate a strategy and explain each error
run a backtest on the QuantPilot engine
show backtest results, metrics and orders
list your backtests that are not attached to a strategy or a chat
The client acts as you. Backtests run under your QuantPilot account, on the same engine as backtests in the app.
Tools
These are the tools your AI client gets once QuantPilot is connected. Seven of them only read. One, Run a backtest, creates a backtest in your account; a client that asks before a non-read-only action, such as ChatGPT, asks only for that one.
Tool | What it does | Access |
QuantScript v2 docs | Reads the QuantScript v2 language reference, the backtest guide and the step-by-step procedure the client follows | Read |
List markets | Lists the exchanges, market types (spot, perpetual) and candle timeframes that have data | Read |
Search symbols | Finds a symbol by name, with its market type, maximum leverage and the dates it has candles for | Read |
Validate a strategy | Checks a strategy with the platform validator and returns each error with the documentation that explains it | Read |
Run a backtest | Submits a backtest to the QuantPilot engine under your account | Write |
Get a backtest | Status, metrics and warnings of one of your backtests, and its code when asked for | Read |
List my backtests | Your backtests that are not attached to a strategy or a chat, newest first, with their headline metrics | Read |
Get backtest orders | The orders of a completed backtest, including rejected ones and why they were rejected | Read |
Next step
To connect your AI client, see Use QuantPilot from Claude, ChatGPT and other AI clients.