DataSinking
Full-text financial reports as clean Markdown.
2 followers
Full-text financial reports as clean Markdown.
2 followers
Clean, structured Markdown for listed companies across China, Korea and Japan. Query by symbol with a simple REST API.
2 followers
2 followers
Hey Product Hunt! π
I built DataSinking because I kept hitting the same wall building RAG apps on Asian
equities: annual reports are PDFs on cninfo/DART/EDINET, and turning them into clean
Markdown β tables intact, chapters separated β is a whole pipeline, not an API call.
DataSinking gives you full-text financial reports from China, Korea & Japan as clean
Markdown, over a simple REST API (or an MCP server for AI agents):
- Full text, not just numbers β the MD&A, risk factors and notes that RAG needs.
- Chapter-level access (?section=MD&A) so you pull one section instead of 300 pages.
- FMP-style symbols: 600519.SS (Moutai), 005930.KS (Samsung), 7203.T (Toyota).
- Free key by email, no signup.
Try it:
curl "https://api.datasink.ing/documents?symbol=600519.SS&with_content=1&apikey=YOUR_KEY"
Happy to answer questions β and I'd love feedback on which markets/formats to add next! π
Full-text financial reports as clean Markdown β 6 markets, MCP + REST API
β DataSinking turns official financial disclosures (10-K/10-Q, annual reports, εεΉ΄ε ±, ζδΎ‘θ¨ΌεΈε ±εζΈβ¦) into clean,
β section-addressable Markdown β the full narrative text with tables and notes intact, ready for LLM reading and RAG.
β
β Covers 6 markets / 13 exchanges: US (SEC EDGAR), China (CNINFO), Japan (EDINET), Korea (DART), Taiwan (MOPS), UK
β (FCA ESEF) β 18,680 companies.
β
β - Full text, not key-values β FMP gives you revenue/EPS; we give you the whole report, notes included
β - Section-addressable β pull just MD&A / Risk Factors, not the 200k-word whole thing
β - MCP server β plug into Claude / Cursor, query in plain language
β - Free tier (email for a key) + $31/year
β I built DataSinking after hitting the same wall repeatedly: getting the full text of a non-US report
β (China/Japan/Koreaβ¦) as clean data was either impossible, $10k+/yr, or token-compressed mush. Most APIs sell
β structured numbers; I wanted the narrative β MD&A, risk factors, the notes with their tables β because that's where
β the real analysis lives and what RAG workflows actually need. There's an MCP server so Claude/Cursor can query it
β directly. Free tier is an email-away key. Would love feedback β especially the MCP experience and what markets to
β add next.