Connecting Client

Connecting

Set up Block Scholes MCP in your preferred client, or use the manual configuration for any other MCP-compatible app.

Choose a client

• Claude

• Cursor

• OpenClaw

Add the Block Scholes skill

After connecting, we recommend creating a shared Block Scholes skill (sometimes called a custom instruction or rule file, depending on your client). This gives your AI assistant persistent context about how to use the Block Scholes tools effectively — keeping data requests, backtests, and output formatting consistent across sessions.

Without a skill, your assistant will still work, but you may find yourself repeating preferences (e.g. "use hourly frequency", "show results in a table", "always include the forward price") in every conversation. The skill handles that for you. See the shared Block Scholes Skill for more info.

Other MCP Clients

Block Scholes works with most AI Agents that supports the Model Context Protocol. If your client is not covered in the dedicated guides above, use the manual configuration below.

Server configuration

Add the following to your MCP client configuration:

{
  "mcpServers": {
    "blockscholes": {
      "url": "https://mcp.blockscholes.com/mcp"
    }
  }
}

The exact file location depends on your client — check your client's documentation for where MCP server configs are stored.

Authentication

All connections authenticate via OAuth. When your client first connects to Block Scholes MCP, you will be redirected to sign in or create a Block Scholes account.

No API keys or manual token management are required for MCP connections.

Available tools

Once connected, your client will have access to three tools.

data_retrieval

Query the Block Scholes data universe. Supports:

• Price data — spot, perpetual, futures, and options across multiple exchanges

• Implied volatility — by strike, delta, or moneyness using SVI or spline models

• Rates — perpetual funding rates and annualised futures rates

• Open interest and volumes — across supported asset types

• Time ranges — latest, historical ranges, or specific timestamps

• Frequencies -- 1-minute, or 1-hour intervals

backtest_strategy

Execute strategy backtests with institutional-grade data. Supports:

• multi-leg positions across spot, perpetual, futures, and options

• theoretical or listed pricing modes

• delta hedging via spot or perpetual instruments

• custom entry and exit dates, tenors, strikes, and quantities

• trade-by-trade PnL with hedge attribution

get_instruments

Look up available instruments within the Block Scholes universe. Use it to discover listed expiries, strikes, and option types before constructing queries.

Tips

• Use a model with strong tool-calling support. Claude Opus 4.5+ is recommended.

• Enable code execution if your client supports it.

• Use ISO 8601 dates, for example 2025-01-01T08:00:00Z.

• Submit all trades in a single backtest call where possible.

• If your client supports skills or system prompts, add the shared Block Scholes Skill.

Need help?

If you hit issues connecting your client, contact us through the console or via your existing Block Scholes support channel.