Quant Data is a statistical skill set for index futures and stock markets, specifically packaged for use by AI agents, and it emphasizes reporting measurement results truthfully. The repository can be natively installed into Claude Code (plugin), Codex CLI (plugin marketplace), and Gemini CLI (extension), and also supports any other MCP client.
Purpose To provide AI agents with callable market statistics endpoints covering U.S. stocks and ETFs, futures, Hong Kong stocks, and A-shares, with outputs including intraday type probabilities, Weis volume waves, option max pain, and market maker Gamma exposure. All results come with historical frequencies and failure-case explanations for agents to cite in conversation.
Main capabilities Four API endpoints: intraday type probabilities, Weis volume waves, option max pain, market maker Gamma. The remote MCP server address is https://api.quantdata.uk/mcp，使用 Streamable HTTP. Without a key, one anonymous market tool response can be obtained once per UTC day per source. Agents can call quantdata_request_free_api_key and pass in an email to apply for a free key; the key is sent only to that email, with no login or credit card required, and the free quota is 10 successful calls per day shared across the four market tools.
How to use Installation: the repository includes installation methods for the Claude Code plugin, Codex plugin marketplace, and Gemini CLI extension. You can also manually copy the SKILL.md file under quantdata/skills/ into the .claude/skills/ directory. For clients without a plugin system, copy the content below the YAML header in SKILL.md into the system prompt or custom instructions, and replace $QUANTDATA_API_KEY with the platform's key storage method. Known working platforms include ChatGPT custom GPTs, Cursor, Windsurf, Cline, GitHub Copilot instructions, n8n, Dify, CrewAI, AutoGen, LangChain, and tools built directly on the OpenAI or Anthropic SDK. Any tool that can send one HTTP request with custom headers can support it, with no SDK, OAuth flow, or callback URL required. Key management: copy the qd_ key from the email into key storage; do not paste it into SKILL.md. After a free key reaches its daily quota, the API returns 429 along with a reset time, and provides a checkout link for manual handling; agents should not loop requests or replace the data source. The paid plan is $149 per month, free for the first 3 days, and checkout is completed self-service at https://quantdata.uk/pricing , with a credit card required in advance. Checking out with the same email as the free key upgrades the key in place. You can cancel at any time in the Stripe customer portal.
Use cases Suitable for having AI agents query market statistics for index futures and stocks, for example asking whether ES=F has broken out of a range and the historical failure frequency of similar breakouts, whether a pullback occurred on low volume, a stock's max pain this Friday, or a stock's market maker Gamma bias. It is also suitable for integrating statistical results into custom GPTs, Cursor rules, n8n workflows, and multi-agent frameworks.
Limitations All outputs are for educational analysis and developer tools only, and do not constitute investment advice, buy or sell recommendations, or price forecasts. Historical performance does not guarantee future results. Signals are unconfirmed on cryptocurrencies and gold, chart pattern test AUC is only 0.54, and Asian session tests return NO_GO. In terms of accuracy, event edge is 52–56%, intraday type top-1 accuracy is 66%, and the majority-class baseline is 37%. Base rates and models are measured on S&P 500 E-mini futures, and readings for individual stocks, gold, or cryptocurrencies are descriptive only and unverified. Max pain is an arithmetic result, and Gamma is a Black-Scholes estimate based on assumptions about unobservable market maker positions; the skill presents them as numbers and estimates respectively. Paid purchases require manual 3-D Secure authentication, and agents cannot purchase on your behalf.
This project is provided by quantdata.uk and is unrelated to quantdata.us; keys are not interchangeable. Free pages include https://quantdata.uk/max-pain and https://quantdata.uk/gex，无需密钥。方法论页面 https://quantdata.uk/methodology Free and public, no key required.
