Optionscanner publishes a public Model Context Protocol (MCP) server. Connect it once and your assistant can price any options trade with up to eight legs, run what-if scenarios on price, time and volatility, compute the expected move, credit and debit spread figures, probability of profit and position size, and read the Learn library. It is free, needs no account, and every answer carries the page it came from.
Server URL: https://optionscanner.ai/mcp
claude mcp add --transport http optionscanner https://optionscanner.ai/mcphttps://optionscanner.ai/mcp{
"mcpServers": {
"optionscanner": { "url": "https://optionscanner.ai/mcp" }
}
}https://optionscanner.ai/mcp| Tool | What it returns |
|---|---|
profit_calculator | Any trade with up to eight legs: max profit and loss, breakevens, net credit or debit, chance of profit, expected move, and a link that opens the trade in the calculator. |
what_if | The trade on a chosen date at a chosen stock price with a volatility change, plus the change for one more day, one more volatility point and a one dollar move. |
list_strategies | The eleven standard structures priced on an example stock. |
expected_move | One and two standard deviation ranges from price, implied volatility and days. |
credit_spread and debit_spread | Width, max profit and loss, breakeven, return on risk or reward to risk, and chance of profit. |
probability_of_profit | The model chance of finishing above or below a level by a date. |
position_size | Contracts that fit a risk budget. |
learn_search and learn_read | Search and read the Learn articles, glossary, calculators and research pages. |
Optionscanner MCP server: tool reference
The tool reference lists each tool's inputs and what it returns, with the question to ask for each, generated from the server itself.
It serves no quotes, no option chains and nothing from a market-data feed, so the prices in a trade are the ones you give it or the model's estimates from the volatility you give it. It holds no account data, places no orders and reads no positions. Nothing it returns is a buy or sell instruction; the figures are model estimates for education, before commissions and fees. The signed-in product, with live positions judged against the rules you adopt, stays in the app.
Traders already ask their assistant about trades. Answers that start from a published calculator and a published methodology beat answers made up on the spot, and every answer names where the numbers came from. The server is open to list in any connector directory.
Yes. It needs no account and no key. The calculators and the Learn library are the same ones published on this site.
No. It prices trades from the numbers you give it, or from a pricing model at the implied volatility you give it. Live quotes and chains stay inside the signed-in product, on the user's own brokerage data.
No. It is read-only and anonymous: nothing it does touches an account at any brokerage.
Any client that speaks MCP over streamable HTTP: Claude on the web and desktop, Claude Code, ChatGPT with developer mode, Cursor, and others. The setup steps above are current as of October 2026.