A-Share MCP Server

Instructions

Implementation Reference

• src/tools/indices.py:65-72 (handler)

  The MCP tool handler for 'list_industries'. Decorated with @app.tool() for automatic registration and schema inference. Logs the invocation and delegates to the core use case logic via run_tool_with_handling.

  @app.tool()
  def list_industries(date: Optional[str] = None, format: str = "markdown") -> str:
      """List distinct industries for a given date."""
      logger.info("Tool 'list_industries' called date=%s", date or "latest")
      return run_tool_with_handling(
          lambda: fetch_list_industries(active_data_source, date=date, format=format),
          context="list_industries",
      )

• src/use_cases/indices.py:38-47 (helper)

  Core implementation logic for listing industries: fetches all stock industry data, extracts unique sorted industries, and formats as markdown table.

  def fetch_list_industries(data_source: FinancialDataSource, *, date: Optional[str], format: str) -> str:
      validate_output_format(format)
      df = data_source.get_stock_industry(code=None, date=date)
      if df is None or df.empty:
          return "(No data available to display)"
      col = "industry" if "industry" in df.columns else df.columns[-1]
      out = df[[col]].drop_duplicates().sort_values(by=col)
      out = out.rename(columns={col: "industry"})
      meta = {"as_of": date or "latest", "count": int(out.shape[0])}
      return format_table_output(out, format=format, max_rows=out.shape[0], meta=meta)

• mcp_server.py:53-53 (registration)

  Top-level call to register_index_tools, which includes the registration of the 'list_industries' handler via its @app.tool() decorator.

  register_index_tools(app, active_data_source)