options-chain-mcp

Name: options-chain-mcp
Author: blake365

@blake365

代码仓库

关于 options-chain-mcp

Option Chain MCP server

#options #mcp #market-data #claude #mcp-server

基本信息

分类

其他

许可证

MIT

运行时

node

传输方式

stdio

发布者

blake365

配置

使用下面的配置,将此服务器添加到你的 MCP 客户端。

{
  "mcpServers": {
    "options-chain": {
      "command": "node",
      "args": [
        "/Full/Route/to/Folder/options-chain/build/index.js"
      ],
      "env": {
        "token": "your_sandbox_api_token_here"
      }
    }
  }
}

工具

未检测到工具

工具是从 README 中自动提取的。维护者可以在 ## Tools 标题下列出工具,即可填充这部分内容。

概览

What is options-chain-mcp?

A read-only Model Context Protocol (MCP) server for options research. It provides AI assistants with tools to query options chains, expirations, quotes, and historical prices, using a pluggable data provider (currently supports Tradier and Alpaca). The server is designed for informational purposes only and cannot place trades.

How to use options-chain-mcp?

Install Node.js, clone the repository, and configure environment variables for your chosen provider (Tradier token or Alpaca API keys). For local use, add the server to Claude Desktop’s configuration file (claude_desktop_config.json). For remote use, deploy on Cloudflare Workers with OAuth 2.1 and connect Claude.ai via custom connector. The server exposes four tools: find-options-chain, find-option-expirations, get-quote, and historical-prices.

Key features of options-chain-mcp

Read-only; no order placement or trade execution
Pre-filters options chains to reduce token usage (20–40× fewer tokens)
Pluggable data provider (Tradier or Alpaca) with identical tool interface
Supports both local stdio (Claude Desktop) and remote OAuth (Claude.ai web)
Automatically drops dead contracts and trims strikes to significant levels
Server-side filtering mimics human trader reading of chains

Use cases of options-chain-mcp

Researching options chains for a symbol and expiration without trade risk
Comparing options data across providers by switching a single environment variable
Fetching filtered, LLM-friendly option chains for analysis in conversation
Running options research from Claude.ai web or Claude Desktop
Using historical price data for OHLCV bars of stocks or options

FAQ from options-chain-mcp

How does options-chain-mcp differ from the official Tradier or Alpaca MCP servers?

The server pre-filters chains server-side: only strikes within a configurable percentage of spot, with real volume/bid/ask, and a "significant strikes" pass are returned. This reduces token usage by 20–40× compared to raw API calls. It is also read-only (no trading endpoints) and supports remote OAuth deployment.

What are the runtime requirements?

Node.js 18+ is required for local use. For remote deployment, a Cloudflare Workers paid plan with Durable Objects is needed. The server is already built; cloning the repo is sufficient for use.

Where does data come from and is it delayed?

Data comes from the configured provider (Tradier or Alpaca). Free/paper accounts typically deliver 15-minute delayed market data. Tradier sandbox tokens expire; Alpaca paper accounts work indefinitely.

What are the provider-specific differences?

Tradier returns daily volume and open interest natively; Alpaca returns null for volume from the snapshot endpoint and merges open interest from a separate contracts endpoint. Greeks and IV are available from both providers on free tiers.

Is the server read-only by design?

Yes. The server intentionally omits any place_order, close_position, or exercise_options_position tools, making it safe for LLM-assisted research without risk of hallucinated trades.