What problem does it solve? Manually compiling an end-of-day A-share market review requires pulling index performance, breadth statistics, sector rotation, capital flows, L1/L2 microstructure factors, news sentiment, and model signal hit rates from many disconnected sources. This Skill automates the entire pipeline into a fixed-structure Markdown and PDF report with a next-day direction assessment. ## Core Features & Use Cases - Automated data collection: Runs daily_review.py and news_review.py against QuantDB parquet data, PostgreSQL inference signals, and Huntly news enrichment to produce stats JSON and a facts Markdown file. - Structured review report: Renders a 13-section report covering indexes, limit-up ladders, sector rotation, margin/northbound capital, L2 microstructure, news sentiment, model signal hit rates, and a six-dimension next-day direction score with confidence stars. - PDF delivery and archiving: Converts the report to a research-style PDF and files it into the stock reports directory visible in the frontend, plus a chat-ready summary. - Use Case: After market close, ask for a daily review of 2026-08-14; the Skill runs the data scripts, writes the report, generates the PDF, and replies with a one-screen market snapshot including tomorrow's direction call. ## Quick Start Ask the assistant to run a daily review for a specific trading day, for example by saying "复盘 20260814", and it will execute the data scripts, write the report, convert it to PDF, and reply with a market summary.
