kbwenLab

File to Markdown

Convert docs to clean Markdown for AI. 100% in-browser.

100% local — your files never leave this browser

Word Excel CSV HTML PDF TXT

Drop files, click to pick, or Ctrl+V to paste

Word (.docx) · Excel (.xlsx) · CSV · HTML · Text PDF · TXT

Try examples:
Markdown
Converted Markdown will appear here
0
tokens (est.)
TRIM

How it works

Each format is handled by a dedicated parser running entirely in your browser: Word (.docx) via Mammoth, HTML/paste via Turndown, Excel/CSV via SheetJS, text-based PDF via pdf.js. Each parser loads only when you first use that format — first paint stays light.

The AI context meter shows an estimated token count — roughly chars÷4 for English, higher per character for CJK. It's labelled as an estimate, useful for gauging whether to trim, not for exact billing. Trim switches strip headers/footers, image data, and blank lines, updating the estimate live so you see reductions like "8,200 → 5,100 tokens."

CJK extraction uses smart joining — adjacent CJK characters don't get spurious spaces inserted between them, and full-width punctuation (，。！？「」) is preserved verbatim. v1 supports only text-based (selectable) PDFs; scanned/image PDFs have no extractable text and the tool gives a clear message instead of empty output. Complex multi-column PDFs may degrade gracefully — the tool notes this in the output rather than over-promising.

Quick guide

Use cases, answers, and nearby tools

Compact below-tool notes that help first-run users and repeated visitors move faster without changing the main interface.

Chinese search: pdf 轉 markdown、word 轉 markdown、文件轉 markdown 免上傳、excel 轉 markdown 表格、把檔案轉成 markdown 餵 ai

Japanese search: pdf markdown 変換、word markdown 変換、ファイル markdown 変換、markdown 変換 アップロード不要

How to use

Run a clean first pass

  1. Drop a file, click to pick, or paste rich text (Ctrl+V) — supports Word / Excel / CSV / HTML / text-based PDF / TXT.
  2. Markdown appears instantly in the output panel. Hit Copy Markdown or Download .md to paste into ChatGPT / Claude.
  3. Use the AI context meter to see the estimated token count, then toggle trim switches to strip headers, images, or blank lines and watch the count drop live.
  4. Click the example presets (CSV, HTML paste) to see a live conversion demo without needing a real file.
  5. Hit Clear to reset and start over — every conversion stays entirely in your browser, leaving no data behind.

Examples

Real jobs this page helps with

  • Student feeds lecture notesDrag a Word handout into the tool, convert it to Markdown in seconds, and paste it into ChatGPT to ask questions — no manual formatting needed.
  • Engineer cuts token costConvert an API reference PDF to Markdown, turn on trim to strip headers and footers, and watch the context drop from 8,000 to 5,000 tokens — saving Claude API cost.
  • Editor converts spreadsheetDrop an Excel file and get a Markdown table ready to paste into Notion or Obsidian — no copy-paste row by row.
  • Japanese PDF extractionConvert a Japanese PDF with full-width punctuation to clean Markdown — no spurious inter-character spaces, no mojibake.
  • CSV to Markdown tableDrop or paste a spreadsheet CSV and get a Markdown table ready to copy into Notion or Obsidian — no row-by-row reformatting.

FAQ

What people usually want to know

Are my files uploaded to a server?

No. All parsing and conversion runs entirely in your browser. File contents are never uploaded or sent to any server. Closing the tab leaves nothing behind.

What formats are supported?

Word (.docx), Excel (.xlsx), CSV, HTML, rich text pasted from Word / Google Docs / web pages, and text-based (selectable-text) PDFs. Each format's parser loads only on first use, so the initial page load stays light.

Can it convert scanned PDFs?

v1 supports only text-based PDFs with selectable text. Scanned (image-only) PDFs have no extractable text — the tool gives a clear message rather than returning empty output. OCR support is deferred to a later phase.

Are tables preserved?

Yes. Excel/CSV converts to Markdown tables. Tables in documents are converted where possible. Complex multi-column layouts may need manual cleanup — the tool notes this in the output rather than over-promising.

How accurate is the token estimate?

It's a heuristic estimate — roughly chars÷4 for English, higher per character for CJK — useful for gauging payload size before pasting into an AI. It's labelled 'estimate' and is not suitable for exact billing.