What problem does it solve? Producing a rigorous, data-grounded deep research report on an individual A-share stock normally requires manually pulling quotes, valuations, capital flows, financials, and news sentiment from many sources and reconciling conflicting analyst views. This Skill automates that entire pipeline using only local QuantMind data, with a strict no-fabrication rule so every number in the report is traceable to a data package. ## Core Features & Use Cases - Multi-Agent Orchestration: Runs 5 parallel analysts (technical, news, capital sentiment, fundamentals, market), then a bull-vs-bear debate, then a research manager who synthesizes a final rated report with target price range. - Local Data Package: The research_data.py script pulls K-line quotes, technical indicators, valuation, L2 capital flow, financial statements, sector context, FinBERT news sentiment, and model inference scores into one JSON file. - Report Delivery: Converts the final Markdown report to PDF and files it under the A-share market reports directory for the stock. - Use Case: Ask to "deep research 600519" and receive a full investment research report with rating, target range, top risks, and tracking signals, all sourced from local QuantDB and news enrichment data. ## Quick Start Ask the assistant to run a deep research analysis on a specific A-share stock such as 600036 or 招商银行 and wait for the multi-agent report with PDF output.
