Claude for Financial Services
anthropics/financial-servicesAnthropic's reference agents, skills and data connectors for banking, research, fund admin and KYC, usable as Claude plugins or Managed Agents.
With caveats
Check first Depth 3/5
Wind AliceMarket Claude skills
Chinese-language finance skills for A-share and global equity analysis, plus Wind data and agent entry points that need a Wind API key.
Our verdict Recommended with caveats
Broad set of analysis prompts with a few tested scripts, but mostly Chinese-language role-play prompts, and the data skills depend on a Wind API key and were not in the copy we reviewed.
What Wind AliceMarket does
AliceMarket mirrors a monorepo of finance skills published around Wind, the Chinese financial data provider. The README (in Chinese) lists 34 skills in four groups: a discovery meta-skill that recommends other skills, a Wind MCP data skill for stocks, funds, indexes, bonds, announcements, news and macro indicators, an agent entry skill for Wind's Alice analysis agent using a streaming agent protocol, and a set of analysis skills. The analysis skills are prompts that define a role, a stepwise method and an output structure. Examples are A-share main-theme identification by market structure and fund behavior, breakout candidate screening, bull and bear case building, DCF modelling with WACC and sensitivity, earnings analysis, position sizing, moat review, guidance-change impact, and four skills that imitate investor styles such as Buffett and Taleb.
In the copy we reviewed, the data and agent entry skills were not present, only the analysis skills, which average roughly 4 to 5 KB each. Three skills carry Python scripts: a backtest evaluator with tests, a DCF validator and a market-utilities helper. The data and agent skills need a WIND_API_KEY from Wind's AI market site, stored in a config file under your home directory. Prompts are long role-play instructions, written for Chinese-speaking traders and researchers, and several give trade-style guidance.
Good for
- Chinese-language A-share research and daily market recaps
- Structured equity analysis prompts such as DCF and earnings reviews
- Users who already have Wind data access
Not for
- Anyone expecting live data without a Wind account
- Using the trade-style outputs as investment advice
- English-only workflows
What's inside depth 2/5
About 40 skill folders averaging 4-5 KB of prompt with a few reference files and only three script folders, one with tests.
- Skills
- 78
- Reference docs
- 267
- Scripts
- 47
- Tests / evals
- 21
- Templates
- 1
- Avg SKILL.md
- 5.8 KB
Counted from the repository file tree. How we score depth.
Dependencies and costs
- Wind API key (WIND_API_KEY) from Wind's AI market site for the data and Alice agent skills
- Python for the backtest, DCF and market utility scripts
Security review
Review before installingA Chinese-language collection of finance analysis skills that depend on Wind's paid data service: they tell the agent to prompt you to install wind-mcp-skill with an unpinned global npx command and to keep a WIND_API_KEY in a plain-text file in your home directory. The few Python scripts we can see are local calculators; the skills that actually call Wind's API are not in the reviewed copy.
Worth knowing
- Runs remote code at an unpinned versionAbout 30 finance skills tell the agent to prompt the user to run npx skills add Wind-Information-Co-Ltd/wind-skills --skill wind-mcp-skill -g -y (unpinned, global, confirmations skipped) when the Wind data skill is missing; the README uses the same command.skills/dip_buy_decision_skill/SKILL.md:15
- Needs an API key for its own servicewind-mcp-skill and wind-alice need a WIND_API_KEY, stored in plain text at ~/.wind-aifinmarket/config (or WIND_API_KEY env, or config.json beside SKILL.md) for Wind's AI market service.README.md:133
- Changes files outside your projectSetup creates ~/.wind-aifinmarket/config and installs skills globally (-g) for every detected agent; README also mentions a node scripts/cli.mjs open-portal command that opens Wind's developer portal in your browser.README.md:133
- Makes network requestswind-mcp-skill queries Wind's MCP service (aifinmarket.wind.com.cn) and wind-alice uses A2A/SSE streaming to it; dcf-model lists yfinance and requests as dependencies and theme-detector references FINVIZ. Those skills' code is not in the files we have.README.md:202
- Installs packages or toolsdcf-model asks for openpyxl>=3.0.0, yfinance>=0.2.0 and requests>=2.28.0 (unpinned lower bounds) via pip.skills/dcf-model/requirements.txt:4
- Relies on a CLI or package we could not reviewRelies on the npx skills CLI and the wind-mcp-skill/wind-alice skills (including scripts/cli.mjs), which are listed in the README but absent from the reviewed copy.README.md:85
What we checked
The repository (402 files) on Oct 29, 2026. Not a full review: 329 of 402 files were not downloaded (large repository; we reviewed the skill folders); we read 2 of 5 code files; it relies on code outside the repository: wind-mcp-skill, wind-alice, wind-find-finance-skill and most other skills named in the README (about 329 readable files, including their SKILL.md files and cli.mjs, are not in the local copy); Wind's MCP / A2A service and the npx skills CLI; Python packages openpyxl, yfinance, requests (unpinned) and a FINVIZ data source.
NVIDIA SkillSpector static scan: 79 pattern hits (MCP Rug Pull 65, Supply Chain 5, Prompt Injection 3, Agent Snooping 2, Excessive Agency 2, Rogue Agent 1). The 79 hits are mostly repeated 'npx skills' install lines, whitespace padding in README tables, unpinned requirements (openpyxl, yfinance, requests have old advisories but no version is pinned) and the README instructions to write WIND_API_KEY to a config file.
A review, not a certification. Skills change after we check them — read the files before installing.
License
No license file or license metadata found — all rights reserved by default. The prompts mimic named investors, which may raise their own rights questions.
Can it run unattended?
Not really. The skills produce reports and trade-style commentary on request and have no loops or verification gates; the market-hours recaps depend on a person acting on them.
Alternatives to Wind AliceMarket
Edgar Change Interpreter
cmdrvl/edgar-change-interpreterA short skill that makes Claude compare SEC 10-K, 10-Q and 8-K filings for material changes and silent risks, with quoted evidence and fixed output sections.
With caveats
Check first Depth 2/5
FAQ
What is Wind AliceMarket?
It is a collection of Chinese-language finance skills tied to Wind's financial data platform. The analysis skills cover A-share themes, DCF, earnings, position sizing and investor-style thinking, and the Wind data skills connect through MCP with an API key.
How do I use the Wind data skills?
According to the README you get an API key from Wind's AI market site and save it as WIND_API_KEY in an environment variable or in ~/.wind-aifinmarket/config. Without a key the discovery skill and the pure-prompt analysis skills still work.
Are the trading skills financial advice?
No. They are prompts that make the model write analysis and trade-style commentary, and they do not execute trades. Treat output as a draft to check, since the model can be wrong about data and figures.
Wind AliceMarket vs anthropics/financial-services?
AliceMarket is oriented to Chinese markets and Wind data in Chinese. Anthropic's financial-services repository is a set of plugins for general finance workflows. Choose by market and data source.
Sources: GitHub repository (stars, license, file tree, last push Oct 15, 2026); install counts from skills.sh via Best Skills (CC BY 4.0). Review based on the README and SKILL.md files as published.