MCP Server Documentation

Everything you need to connect your AI assistant to RixLabs backtesting tools via the Model Context Protocol.

What is the RixLabs MCP Server?

The RixLabs MCP server gives AI assistants (Claude, ChatGPT, Cursor, and other MCP-compatible clients) direct access to crypto market backtesting infrastructure. Through natural language, users can:

Write and validate Python trading strategies
Run backtests on supported crypto perpetual futures markets
Fetch performance metrics, trades, and equity curves
Compare multiple strategy variants side-by-side
Explore historical market data availability
Access built-in strategy documentation and patterns

Setup

Prerequisites

A RixLabs account (sign up at rixlabs.ai)
An active free trial or support-approved access
An MCP-compatible AI client (Claude Desktop, ChatGPT, Cursor, Claude Code, etc.)

Claude Desktop / Claude Code

Add the following to your MCP configuration file:

{
  "mcpServers": {
    "rixlabs": {
      "url": "https://mcp.rixlabs.ai/mcp"
    }
  }
}

On first connection, you will be prompted to authenticate with your RixLabs account via OAuth.

ChatGPT

RixLabs is available as a connector in the ChatGPT plugin/connector store. Search for "RixLabs" and enable it. Authentication happens automatically via OAuth when you first use a RixLabs tool.

Other MCP Clients

Any client that supports the MCP specification can connect using the HTTP transport:

Server URL: https://mcp.rixlabs.ai/mcp
Transport:  HTTP
Auth:       OAuth 2.0 (initiated on first connection)

Available Tools

The MCP server exposes the following tools to your AI assistant:

Strategy Development

validate_code

Validate Python strategy code against framework rules before submitting. Returns actionable diagnostics.

run_backtest

Submit validated strategy code for backtesting with specified symbol, date range, timeframe, and parameters.

stop_backtest

Cancel a queued or currently running backtest.

Results & Metrics

metrics

Get performance metrics (Sharpe, returns, drawdown, win rate) for a completed backtest.

trades

Retrieve the trade log for a completed backtest run.

mtm

Retrieve mark-to-market (intraday PnL) data for a completed run.

compare_backtests

Compare multiple completed runs side-by-side with correlation analysis.

Data & Status

available_duration

Check the available historical data date range for a symbol and segment.

data

Fetch historical candle data for a symbol over a date range.

day_data

Get single-day spot, futures, expiry, and option chain data.

strategy_status

Check the lifecycle status of a submitted backtest run.

active_backtests

See how many backtests are currently queued or running.

user_plan_info

Check your plan, quota, rate limits, and parallel run capacity.

Analysis Mode

analysis

Retrieve analysis payload (movements and points) from an analysis-mode run.

movement_analysis

Get time-series movement rows recorded by an analysis run.

movement_metrics

Compute statistical metrics from saved movement analysis data.

point_analysis

Get discrete point rows recorded by an analysis run.

point_metrics

Compute statistical summaries from saved point analysis data.

Documentation

list_skills

List all available strategy documentation topics.

get_skill

Read a specific strategy documentation topic (API reference, patterns, indicators, etc.).

Supported Markets

Crypto Perpetual Futures

Symbol	Market Type	Notes
BTCUSDT	Perpetual futures	24/7 UTC market data
ETHUSDT	Perpetual futures	Historical research and backtesting
SOLUSDT and more	Perpetual futures	Use available_duration to confirm coverage

Example Prompts

Strategy Development

"Write a BTCUSDT breakout strategy that trades only during high-volume sessions. Validate it and run it from Jan to Mar 2025."

Performance Analysis

"Give me the Sharpe ratio, max drawdown, and win rate for my last backtest. How does it compare to the straddle-v1 run?"

Market Research

"What date range is available for SOLUSDT? Show me the daily range and volatility profile for March 2025."

Troubleshooting

Authentication fails or token expired

Disconnect and reconnect the RixLabs MCP server in your client settings. This re-initiates the OAuth flow and refreshes your token.

Backtest stuck in "queued" state

Check your active backtest count with active_backtests. Free plans have a limit on parallel runs. Wait for a current run to finish, or cancel one with stop_backtest.

Validation errors in strategy code

The validate_code tool returns specific diagnostics. Common issues: import statements in code (not allowed), module-level variables, using qty instead of lots, or non-lambda data inputs to indicators. Ask your AI assistant to fix the listed diagnostics.

No data available for requested date range

Use available_duration to check what historical data exists for your symbol. The requested start/end dates must fall within the available window.

Connection issues with SSE transport

Ensure your client supports SSE (Server-Sent Events) transport. If behind a corporate proxy, SSE connections may be blocked. Try connecting from a different network.

Rate Limits & Quotas

The free trial is available once for the first 30 days after signup. Use the user_plan_info tool to check your current quota at any time.

Parallel backtests are capped at 2 during the trial
The trial includes up to 1000 metered tool-call units
API rate limiting applies across all MCP tool calls
Complete available data is included during the trial

Support

If you encounter issues not covered above: