guangxiangdebizi/FinanceMCP

What it solves

FinanceMCP provides AI Agents with a unified, routable, and traceable way to access financial data across multiple markets. It eliminates the need for developers to manually integrate multiple disparate financial APIs by offering a standardized set of tools that can be plugged into MCP-compatible clients like Claude, Cursor, and Codex.

How it works

It acts as an MCP (Model Context Protocol) server built with Node.js and Express. It exposes 19 stable tools that route requests to various data providers based on configured priorities. It supports two transport modes: local stdio for direct integration and Streamable HTTP for remote access. The system features an automatic fallback mechanism where if a primary data source (e.g., Tushare) is unavailable or doesn't cover the requested data, it tries the next provider in the priority list.

Who it’s for

This project is designed for developers building AI Agents for financial research, investment analysis, and intelligent document management (such as the FinNote system), as well as users of MCP-compatible AI editors and assistants.

Highlights

• Multi-Market Coverage: Access to A-shares, Hong Kong stocks, US stocks, indices, funds, bonds, futures, forex, macroeconomics, and crypto assets.
• Intelligent Routing: Dynamic routing between Tushare, Qveris, Twingly, and Binance with automatic fallback and source transparency.
• Technical Indicator Engine: Built-in calculation for MACD, RSI, KDJ, BOLL, and MA, with automatic historical data pre-fetching.
• Flexible Deployment: Supports both local stdio and remote HTTP endpoints with request-level credential isolation.