A-Share Financial Statements via AI Agents: Income, Metrics & Dividends
Doing A-share fundamental research? Let an AI agent pull income statements, financial metrics and dividend data in one go, then screen stocks by your own thresholds instead of reading reports one by one.

Tool names in this guide are for reference only and may have changed or been renamed. Please use the latest tools in the tool marketplace when implementing.
Key takeaways
- Have an agent pull income statements, financial metrics and dividends in one request
- Replace reading reports one by one with batch screening across a basket of stocks
- One connection and one key feed financial data straight into the agent for computation
Tools used
Why A-share financial statements are a grind
Financial statements are the core of fundamental research, yet reading even one company's income statement, financial metrics and dividend history is tedious — and comparing dozens of firms across a sector makes the workload explode. You open report after report, copy numbers into a spreadsheet, normalize them into ROE, gross margin and growth rates, and only then get to the actual judgment. Data wrangling eats most of the time; the analysis gets squeezed to the end.
If you're after an efficient way to work with A-share financial statements, the real need usually isn't "can I look up one company" — it's "can I pull data in bulk and have the tool compute and screen for me."
What the workflow looks like
Here's one full fundamental screen abstracted into a repeatable loop:
- Give the agent a ticker or a sector scope.
- Pull revenue, profit and growth from the income statement.
- Pull ROE, gross margin, leverage and other derived metrics.
- Check dividend history and yield.
- Have the agent screen a basket of stocks against your thresholds.
Throughout, you only issue natural-language instructions — fetching, normalizing and computing are all handled by the agent.
How each step works
Step 1: Define the scope
Tell the agent exactly what you want. use ae to analyze the latest financials of the largest companies in a given sector. The agent breaks the scope into concrete tickers and starts pulling data.
Step 2: Pull the income statement
Use an income-statement tool (such as ae_tushare_income) to fetch revenue, operating profit and net income, with reporting periods aligned automatically. This is the base for judging whether a company earns and grows.
Step 3: Pull financial metrics
Use a financial-metrics tool (such as ae_tushare_fina_indicator) to retrieve ROE, gross margin, net margin, debt-to-asset ratio, earnings per share and other derived figures in one shot — exactly the quantitative fields screens rely on, without manual conversion.
Step 4: Check dividends
Use a dividend tool (such as ae_tushare_dividend) to pull dividend history and yield, which speak to the quality and sustainability of shareholder returns. Income-oriented strategies lean heavily on this step.
Step 5: Screen by metrics
Hand the agent your thresholds, e.g. use ae to find tickers with ROE above 15% for three straight years, debt-to-asset below 60%, and consistent recent dividends. The agent computes over the data it pulled and filters it, handing back a candidate list directly.
| Approach | Reading reports by hand | Agent batch screen |
|---|---|---|
| Coverage | One company at a time | A whole sector / basket at once |
| Data retrieval | Open reports, copy numbers | Tool returns structured data |
| Metric computation | Done yourself in a spreadsheet | Agent normalizes and computes |
| Screening | Manual, line by line | Filtered against thresholds in one pass |
| Output | A hand-built spreadsheet | A repeatable candidate list |
How to get started
- Sign up — new users get 1,000 free credits.
- Connect AgentEarth to your agent — in-app search in WorkBuddy / Trae, or an MCP config line for Claude Code / Cursor / VSCode.
- Ask in natural language and let fetching, computing and screening flow in one go.
FAQ
Q: Which financial data is supported? Income statements, financial metrics and dividends are all available, covering revenue and profit, derived figures like ROE and gross margin, and dividend and yield data — enough to drive a full fundamental screen.
Q: Can it screen stocks by metrics directly? Yes. State your thresholds in natural language and the agent computes over the retrieved data and filters it, returning a candidate list rather than just a pile of raw numbers.
Q: How much history does it cover? It supports multiple historical reporting periods so you can see growth trends and dividend consistency; the exact range depends on the ticker and data availability.
