Public Opinion Risk Warning

Problem

When users need to assess whether a stock, company, fund, sector, or portfolio has recently attracted negative public attention, relevant signals are often scattered across litigation, regulatory penalties, executive changes, compliance issues, and media reports. Public Opinion Risk Warning targets this workflow by turning raw news search into a structured risk assessment process and producing typed reports that can be reviewed before investment, risk, or research decisions.

How it works

The skill relies on the Hangsheng Juyuan MCP database and is constrained to database tools rather than browsers or crawlers. A typical run starts by identifying the target type, then applies different evaluation criteria for companies/stocks, funds/fund managers, industry sectors, and portfolios. It searches around a keyword using parallel news and public opinion tools, prioritizing the last 1 month of data and extending to 2 months only when needed. Results are ranked by relevance, deduplicated across sources, and filtered to exclude research report content.

• Companies/stocks: focuses on delisting, investigations, major litigation, large penalties, executive changes, and debt default.
• Funds/managers: focuses on liquidation, compliance violations, manager changes, large drawdowns, abnormal size changes, and rating downgrades.
• Sectors: focuses on policy shocks, regulatory actions, supply-chain disruptions, demand declines, overcapacity, and technology substitution.
• Portfolios: focuses on concentration, sector correlation, risk contagion, and overall exposure.

The output is selected by target type: company/stock reports, fund reports, sector briefs, and portfolio diagnostics. Reports also classify the communication stage, such as emergence, fermentation, outbreak, or decline, and cross-check fundamentals, industry conditions, and market sentiment.

Boundaries

The skill only handles public-opinion risk for these four target types and does not answer general chat. Outputs must be grounded in Hangsheng Juyuan database results and must not use browser scraping, competitor data, or research report views. If no valid negative items are found in the last two months, it returns a clear statement that no major negative public-opinion risk was detected rather than guessing. Users should provide a clear company name/code, fund name/code, sector name, or holding list to reduce ambiguity.