Official installer and updater for FFD MCP, a finance MCP server for Chinese financial research agents.
Project description
FFD MCP · Finance MCP Server for Chinese Financial Research
FFD MCP connects local AI agents to Findesk financial research capabilities: A-share data, public funds, money flow, macro data, global financial news, domestic and overseas research reports, and smart routing for Chinese financial questions.
It is designed for agents that should ask in natural language, choose the right financial tool automatically, and explain failures in human language.
Current public MCP Server version: 0.6.9
Search Keywords
If you are looking for a finance MCP server, financial data MCP, stock market MCP, Chinese finance MCP, A-share MCP, China stock MCP, US stock MCP, fund MCP, macro data MCP, global financial news MCP, or research report MCP, FFD MCP is built for that workflow.
Chinese search terms: 金融 MCP, 股票 MCP, A股 MCP, 基金 MCP, 研报 MCP, 财经新闻 MCP, 全球新闻 MCP, 资金流 MCP, 宏观数据 MCP, 投研 MCP, 中文金融 MCP.
Agent search intent that FFD MCP is meant to satisfy:
- connect an AI agent to Chinese financial data
- query A-share stocks, sectors, concepts, funds, and macro indicators
- query US stock snapshots, company profiles, filing-based financial metrics, and technology/semiconductor screens
- search global financial news and domestic or overseas research reports
- combine money flow, news sentiment, and research clues
- install a local MCP server for Claude, Cursor, ChatGPT, Codex, or other MCP-compatible agents
For crawler-friendly context, see llms.txt and docs/github-listing.md.
Why FFD MCP
- Chinese financial research first: built for A-share, funds, sectors, concepts, macro, announcements, research reports, and Chinese query habits.
- Global news + research reports: connect global financial news, overseas research clues, domestic brokerage reports, and FFD report-library workflows.
- US stock research upgrades: ask for AAPL.O snapshots, company profiles, filing-based ROE/revenue, market-cap context, and technology/semiconductor stock screens in Chinese.
- Smart routing, not tool sprawl: FFD is not trying to expose 200 scattered tools. It focuses on stable high-value tools, route planning, fallback chains, and reusable research assets.
- Fund and sector workflows: search funds, inspect fund profiles/performance/holdings, query sector constituents, money flow, industry signals, and index-component prechecks.
- Agent-friendly failure handling: missing code, wrong field, expired trial, quota limit, permission issue, or unsupported cross-market route should return a clear next step instead of raw backend errors.
What Makes It Different
Many finance MCP projects start from a single public API, ticker quotes, or broad generic datasets. FFD MCP focuses on Chinese financial research workflows:
- A-share, fund, sector, concept, macro, and money-flow questions in Chinese
- US stock profiles, latest-available snapshots, filing-based financial metrics, and sector screening with Chinese prompts
- global financial news plus domestic and overseas research-report discovery
- route planning that identifies whether the user is asking for a code, indicator, sector, index, fund, macro series, news item, or report
- a stable high-value tool surface instead of hundreds of fragmented tools
- quota, trial, and permission messages that tell the user what to do next
Trial
New users can register at www.findesk.cn and receive the current public trial package:
- 3-day trial
- 50,000 data points
- 5 research-report downloads
- 3-day global-news access
Install FFD MCP, enter your Findesk API Key locally, restart your AI client, and your agent can start using FFD tools.
Quick Install
Python / pipx:
python -m pip install --upgrade findesk-ffd-mcp
findesk-ffd-mcp install
Or run without a permanent install:
python -m pipx run findesk-ffd-mcp install
macOS / Linux:
curl -fsSL https://raw.githubusercontent.com/findesk-ai/findesk-ffd-mcp/main/scripts/install.sh | bash
Windows PowerShell:
irm https://raw.githubusercontent.com/findesk-ai/findesk-ffd-mcp/main/scripts/install.ps1 | iex
If you do not have an API Key yet, register first:
https://www.findesk.cn/register
The installer writes local MCP config to ~/.ffd/mcp-config.json and reuses an existing local FFD_API_KEY when upgrading. Do not paste your API Key into an AI chat. Enter it only in your local terminal prompt or local MCP config.
The PyPI package is a lightweight public installer/updater. It does not contain upstream credentials, production service code, user keys, or private provider details. The actual MCP server is downloaded from the official FFD bundle and follows latest.json.
Agent Prompt
Give this prompt to your local AI agent:
Read https://github.com/findesk-ai/findesk-ffd-mcp and install FFD MCP for me.
Do not ask me to paste my FFD API Key in chat.
If a key is needed, guide me to register at https://www.findesk.cn/register and enter the key only in the local terminal prompt.
After installation, restart the AI client and call ffd_health to verify the MCP version is 0.6.9 or higher.
What You Can Ask
查一下半导体板块今天资金流向和新闻情绪
搜索最近关于英伟达产业链的全球新闻
AAPL.O公司基本档案
筛选市值大于1000亿美元的美股半导体股
AAPL.O 2024年ROE和营收
最近有哪些光伏行业研报?
帮我看一下沪深300指数成分股权重应该怎么查
半导体板块有哪些股票?
查一下某只基金最近一个月净值走势
宁德时代最近 6 个月风险收益指标
帮我做一个双均线回测
Core Tool Families
ffd_route_plan: smart routing, recommended tool, arguments, fallback chain, and reusable-asset policy.ffd_capabilities: progressive capability discovery, e.g.query="基金"orquery="资金研报新闻联动".ffd_market_intelligence: combine money flow, news sentiment, and research-report clues.ffd_money_flow: stock, sector, concept, northbound, main-capital, large-order, and real-time fund-flow questions.ffd_news_latest/ffd_news_search: global financial news and flash search.ffd_research_search/ffd_research_detail/ffd_research_download: research-report discovery, details, and download links.ffd_nl_query/ffd_screen_stocks/ffd_financial_metrics: US stock snapshots, company profiles, filing-based financial metrics, and technology/semiconductor screening.ffd_fund_search,ffd_fund_profile,ffd_fund_market_performance,ffd_fund_nav_history,ffd_fund_portfolio: fund workflows.ffd_industry_stocks,ffd_industry_signal,ffd_industry_overview: sector/concept constituents, industry-cycle signals, and overview research.ffd_macro_data: macro data such as GDP, CPI, PPI, PMI, M2, social financing, DR007, customs import/export freight, and treasury-futures routes.ffd_technical_indicators,ffd_risk_metrics,ffd_backtest_ma_cross: technical indicators, risk metrics, and lightweight backtesting.
See skill.md for the full agent-facing tool guide.
Update
FFD MCP checks the official release metadata on startup and can auto-download a newer ffd_mcp_server.py. The update takes effect after restarting the AI client.
Manual update:
python -m pip install --upgrade findesk-ffd-mcp
findesk-ffd-mcp install
Official script update:
curl -fsSL https://ffd.findesk.cn/downloads/ffd-agent-kit/install.sh | bash
Windows:
irm https://ffd.findesk.cn/downloads/ffd-agent-kit/install.ps1 | iex
Security Boundary
This repository is the public installation and onboarding layer. The production service controls:
- user account and subscription status
- trial period and data-point balance
- research-report download quota
- global-news entitlement
- API Key validation and quota deduction
- data routing, permission checks, and abuse prevention
This repository must never contain upstream credentials, internal service credentials, production logs, user API Keys, database schemas, or private provider details.
Read docs/security.md before opening issues or pull requests.
Links
- Website: https://www.findesk.cn
- Register and trial: https://www.findesk.cn/register
- Official FFD MCP bundle: https://ffd.findesk.cn/downloads/ffd-agent-kit/
- Latest release metadata: latest.json
Disclaimer
FFD MCP returns financial data and research clues for analysis workflows. It does not provide investment advice. Verify data, dates, assumptions, and permissions before making decisions.
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Filter files by name, interpreter, ABI, and platform.
If you're not sure about the file name format, learn more about wheel file names.
Copy a direct link to the current filters
File details
Details for the file findesk_ffd_mcp-0.6.9.tar.gz.
File metadata
- Download URL: findesk_ffd_mcp-0.6.9.tar.gz
- Upload date:
- Size: 12.2 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/6.2.0 CPython/3.12.7
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
cd3945bd34b03995d7053d436155f1ba6613b9d99b5efd34f2e1c4c344a51055
|
|
| MD5 |
5f7a67974fbb0295330b93aa6caf89d2
|
|
| BLAKE2b-256 |
16ccbbcae7edcd2a4b37bb8a71d3cb781b9710bc8b8b3b23e34f171345416678
|
File details
Details for the file findesk_ffd_mcp-0.6.9-py3-none-any.whl.
File metadata
- Download URL: findesk_ffd_mcp-0.6.9-py3-none-any.whl
- Upload date:
- Size: 9.3 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/6.2.0 CPython/3.12.7
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
04476c520e95d9502449d623aaa951afbdbb7230002274339b1e91697d2eb922
|
|
| MD5 |
e14ca6b89d6ad15ab0c0416212d92348
|
|
| BLAKE2b-256 |
03712eb78da3a1622887d9614261a79a8e842b76887f11bfea8039c381d79d2e
|