CPA Data Hub

Problem

When users ask questions like how many times a firm was penalized, how a company's audit opinions changed, or how many annual report inquiries were issued in 2024, Agents often lack a stable structured source. CPA Data Hub targets China audit market queries with public REST endpoints for audit opinions, firm penalties, company penalties, annual report inquiries, and stats.

How It Works

The skill does not require an MCP server. It calls mycpacp.com endpoints using curl. A typical flow is:
- query list endpoints with fields such as firm_name, company_name, ts_code, audit_agency, end_date, and report_year;
- use the returned id to fetch detail endpoints for full text in audit_text, md_content, or md_text;
- prefer /api/skill/stats/ for broad trends instead of paging through many results.

The data includes useful fields such as audit_result, audit_fees, audit_sign, fine_amount, source_office, penalty_type, exchange, and report_year. keyword searches full text but is more expensive, so exact fields are preferred.

Limits and Notes

• The endpoints are login-free and require no API key, but /api/ requests are limited to 10r/s; batch paging should be serial with intervals.
• List endpoints omit large fields; call detail endpoints only when the user explicitly needs full text.
• keyword must be at least 2 characters and may return truncated: true beyond 200 results.
• audit_fees is numeric in CNY; outputs should show company names first, convert large amounts, and avoid exposing endpoint paths, parameter names, or curl commands.