A-Share MCP Server

Community

MCP server for China A-share market data and financials via Baostock

How to install ↓ Compare GitHub source repository ↗

Category: Other #42 of 230
Stars: ★ 646 Popular
Transport: stdio (local process)
Runtime: Python 3.10+ · Prebuilt binary uv package manager
Credentials: No credential needed
License: MIT
Last commit: 10 months ago ⚠
Tools: 41

64FMRS · C

This server offers broad coverage (41 tools) for China A-share market, financial, and macroeconomic data queries. Setup requires manually specifying absolute paths and the project was primarily developed/validated on Windows, making it suitable for individual developers and research use, but it lacks authentication and depends on a free third-party data source, so it is not suited for production environments requiring strict data reliability or timeliness.

Strongest · Documentation 16/20 Weakest · Reliability 7/20

Reliability

7/20

Security and permissions

15/20

Maintenance

12/20

Documentation

16/20

Setup experience

14/20

Why each score

Reliability 7/20: The README lists 41 tools with clear categorization, but the repository provides no verifiable test suite, CI configuration, or documented error-handling behavior for cases like missing data, API timeouts, or bad parameters. Since this is a static review with no execution and no test/CI evidence present, the score is capped at 12 per the calibration rule; given the plausible but entirely unverified nature of the claims, a 7 is assigned.
Security and permissions 15/20: The server relies on the free Baostock data source and exposes only read-style market/financial data query tools (K-line, financial statements, macro data, etc.), with no evidence of credential storage, financial transactions, file writes, or irreversible destructive actions, and no apparent need for elevated network egress — so no red-line risk is present. However, the documentation does not explicitly describe permission scoping, data boundaries, input validation, or abuse safeguards, nor does it discuss rate limiting or third-party data compliance, so it falls short of full marks; 15 is given, reflecting incomplete disclosure of scoping/permissions rather than any concrete risk.
Maintenance 12/20: The project carries an MIT license, has meaningful community traction (646 stars, 17 open issues, not archived), and the README documents a recent update (2025-12-25, adding the get_fina_indicator tool), indicating some ongoing maintenance. However, there is no visible release history, commit cadence, issue-response SLA, dependency update policy, or security contact channel, leaving governance and versioning largely undocumented; 12 is given.
Documentation 16/20: The README is fairly comprehensive, covering project structure, prerequisites, Baostock's data update schedule, installation steps (uv venv/uv sync), concrete configuration examples for two mainstream clients (JSON-based IDEs and CherryStudio), and a full categorized list of the 41 tools with Chinese descriptions. It lacks per-tool parameter/return-format documentation, rate-limit details, error codes, or a dedicated troubleshooting section, and non-Windows platform details are sparse; 16 is given.
Setup experience 14/20: Setup steps are clear and short (create a virtual environment, uv sync, then write a JSON or CherryStudio config), with concrete config examples and path-escaping notes that lower the barrier to entry. However, the project was explicitly developed and apparently tested only on Windows ('本项目于 Windows 环境下开发'), with no full verification evidence for macOS/Linux, and there is no automated install script, CI smoke test, or execution evidence to raise confidence; per the static calibration rule this dimension cannot exceed 15, so 14 is given.

Static review · not runListed 2026-08-26

Read the FMRS scoring method →

Fit and risk

What it can accessUses the network

Best for

Developers who need programmatic access to A-share market and financial data
Users doing A-share Q&A or research/analysis through an AI assistant
Individuals wanting free A-share data access without a paid data account

Not for

High-frequency trading scenarios needing real-time tick-level or order-book data
Users needing Hong Kong, US, or other non-A-share market data
Institutional production environments with strict requirements on data accuracy/timeliness (data source is the free, unofficial Baostock service)

Required permissions

Runs a local Python process launched via uv
Outbound network access to fetch market and financial data from Baostock
No API keys or account credentials required

Risks and side effects

Data comes from the free third-party source Baostock, not officially guaranteed, with update delays (e.g. daily K-line data isn't stored until after 17:30)
The project was primarily developed and tested on Windows; cross-platform (macOS/Linux) compatibility is not thoroughly verified
No authentication mechanism — any client with access to the MCP server can call all tools and query all data
If Baostock's underlying interface changes, some tools may break without official support

Setup

Before you start

Runtime：Python 3.10+ · Prebuilt binary uv package manager

1) Ensure Python 3.10+ and the uv package manager are installed; 2) Clone the repo and run uv venv in the project root to create a virtual environment; 3) Activate it (Windows: .venv\Scripts\activate; macOS/Linux: source .venv/bin/activate); 4) Run uv sync to install dependencies; 5) In an MCP-capable client (Cursor, VS Code, Trae, CherryStudio, etc.), add a stdio MCP server entry with command set to uv (or its absolute path) and args set to --directory, the absolute path to the project root, run, python, mcp_server.py; 6) If the client reports ModuleNotFoundError, also set workingDirectory to the project root's absolute path.

claude_desktop_config.json

{"mcpServers":{"a-share-mcp":{"command":"uv","args":["--directory","/path/to/a_share_mcp","run","python","mcp_server.py"],"transport":"stdio"}}}

Shown for Claude Desktop. Other clients may use a different file or key (VS Code uses "servers") — the configurator below converts it.

.vscode/mcp.json

{
  "servers": {
    "a-share-mcp": {
      "command": "uv",
      "args": [
        "--directory",
        "/path/to/a_share_mcp",
        "run",
        "python",
        "mcp_server.py"
      ],
      "transport": "stdio"
    }
  }
}

Goes in your project's .vscode/mcp.json (VS Code uses a "servers" key).

Terminal

claude mcp add a-share-mcp -- uv --directory /path/to/a_share_mcp run python mcp_server.py

Run it in a terminal; replace any <…> placeholders with your own values first.

Generate and validate client config

Check that it works

After configuring the client, check that tools like get_historical_k_data appear in the tool list; or ask 'Is today a trading day?' — a result from is_trading_day confirms the connection works.

Troubleshooting

On ModuleNotFoundError, verify `uv sync` was run and try explicitly setting workingDirectory to the project root in the client config
Confirm the uv path in `command` is accessible/executable by the client; use the absolute path to uv.exe if needed
On Windows, escape paths with double backslashes (e.g. `C:\\Users\\...`); macOS/Linux use forward slashes with no escaping needed
If today's data is missing, check Baostock's official data ingestion schedule (e.g. daily K-line after 17:30, minute K-line the next day after 11:00)
In CherryStudio, if the enable toggle appears unresponsive after configuration, navigate to another menu and back to the MCP settings page to refresh the status

Things to try

Once connected, you can ask your AI assistant things like:

Fetch the last 30 days of historical K-line data for Kweichow Moutai
List all constituents of the CSI 300 index
Show me the financial indicator summary for Ping An Bank this quarter
Is today a trading day? What is the nearest trading day?

Tools 41

get_historical_k_data read-only

get_stock_basic_info read-only

get_dividend_data read-only

get_adjust_factor_data read-only

get_profit_data read-only

get_operation_data read-only

get_growth_data read-only

get_balance_data read-only

Show 33 more tools

get_cash_flow_data read-only

get_dupont_data read-only

get_performance_express_report read-only

get_forecast_report read-only

get_fina_indicator read-only

get_trade_dates read-only

get_all_stock read-only

search_stocks read-only

get_suspensions read-only

get_stock_industry read-only

get_index_constituents read-only

get_sz50_stocks read-only

get_hs300_stocks read-only

get_zz500_stocks read-only

list_industries read-only

get_industry_members read-only

get_deposit_rate_data read-only

get_loan_rate_data read-only

get_required_reserve_ratio_data read-only

get_money_supply_data_month read-only

get_money_supply_data_year read-only

get_latest_trading_date read-only

get_market_analysis_timeframe read-only

is_trading_day read-only

previous_trading_day read-only

next_trading_day read-only

get_last_n_trading_days read-only

get_recent_trading_range read-only

get_month_end_trading_dates read-only

get_stock_analysis read-only

normalize_stock_code read-only

normalize_index_code read-only

list_tool_constants read-only

Use cases

Query a stock's historical K-line data, adjustment factors, and dividend/distribution records

Retrieve financial indicators such as profitability, growth, and solvency for listed companies

Look up constituents and industry classification for indices like SSE 50, CSI 300, CSI 500

Query macroeconomic data such as deposit/loan rates and money supply

Perform trading-calendar date calculations (previous/next trading day, last N days, etc.)

Supported clients

Claude DesktopPartial support

Cursor

VS Code

Trae

CherryStudio

Listed from the project's documentation, not tested by this site.

Overview

a-share-mcp is an MCP server focused on the China A-share stock market, built on the free Baostock data source. It exposes 41 tools covering stock basic info, historical K-line (candlestick) data, adjustment factors, dividend records, six categories of financial indicators (profitability, operations, growth, solvency, cash flow, DuPont analysis), performance express/forecast reports, index constituents (SSE 50, CSI 300, CSI 500), industry classification, macroeconomic data (deposit/loan rates, reserve requirement ratio, money supply), and trading-calendar/date utilities. It is written in Python, managed with uv, and communicates with MCP clients over stdio.

Similar servers

Hevy MCP 77 · B

Manage your Hevy workout data from AI assistants via MCP.

★ 517 · Tools 26 Compare with this →

Firecrawl MCP Server 75 · B

Firecrawl's official MCP server for web search, scraping, and structured extraction for AI agents

★ 7.6k · Tools 11 Compare with this →

BioMCP 75 · B

One binary. One grammar. Evidence from the biomedical sources you already trust.

★ 659 · Tools 11 Compare with this →

Compartment 75 · B

Encrypted, fully offline long-term memory for AI agents

★ 583 · Tools 14 Compare with this →

Compare with other MCP servers

Source revision 4a2d4b730e13 Data synced 2026-10-19 Read the FMRS scoring method