Intelligent Financial Report Analysis Assistant

Problem

Financial report analysis often stalls on scattered formats, inconsistent accounting standards, and subjective judgment. Key indicators from PDF statements need manual extraction, while A-share, US-listed, and HK-listed reports can be compared incorrectly if CAS, US GAAP, and IFRS differences are ignored. Non-recurring gains or losses, consolidated statements, seasonal swings, and fraud signals are also easy to miss. This skill codifies a senior financial analyst workflow and performs objective, data-driven analysis without buying or selling advice.

How It Works

The skill accepts a financial report PDF, text file, or link. It first uses scripts/financial-analysis.py to extract key income statement, balance sheet, and cash flow metrics, such as operating revenue, net profit, debt-to-asset ratio, and operating cash flow. It then runs DuPont analysis, decomposing ROE = net margin × asset turnover × equity multiplier, and benchmarks against references/行业财务指标基准.md for comparable companies or industry standards. The final Markdown report follows references/templates/report-template.md, including core conclusions, indicator tables, benchmarking analysis, data sources, and risk warnings.

Boundaries

It is suitable for engineers, analysts, or operators who need a structured read of financial statements, but it is not an investment adviser. For highly seasonal businesses, year-over-year comparisons should be preferred over quarter-over-quarter readings. For corrupted files, encrypted PDFs, or missing cash-flow statements, treat outputs as preliminary and check the anomaly notes. Risk flags should be traced back to source data before being used in audit, financing, or external disclosure contexts.