Markdown Table from Screenshot on Mac

Tables in slides, PDFs, and web UIs are often images. Retyping them into Markdown is slow and error-prone.

ClariRec can detect a regular table layout in an OCR result and let you copy it as a Markdown table — ready for Obsidian, Notion, GitHub, or any Markdown editor.

What you get

  • A Markdown pipe table you can paste directly
  • Optional .md export that prefers the table when one is detected
  • On macOS 26, native document recognition can improve table structure and also expose lists as Markdown lists

This is part of the free OCR workflow. It does not require Pro or a cloud OCR provider when you use System OCR.

Step-by-step

  1. Press ⌥S and select the table region as tightly as you can.
  2. Wait for recognition in Popup mode (easiest for checking structure).
  3. If ClariRec detects a table, use the Table copy action in the result window.
  4. Paste into your notes. Adjust header separators only if the source image was ambiguous.

You can also export to .md; when a table is detected, export prefers Markdown table output over plain paragraphs.

When detection works best

Source Likelihood
Clear grid with visible row/column alignment High
Simple spreadsheet-like screenshots High
Tables with light borders or consistent columns Good
Nested tables, merged cells, or heavy styling Lower — expect cleanup
Multi-column magazine layouts that are not true tables May not qualify as a table

Crop tightly. Including titles, footnotes, or sidebars in the same selection makes structure harder to infer.

macOS 26 native document recognition

On macOS 26, ClariRec can use Apple’s native document recognition request when the setting is enabled (default on, shown only on macOS 26+).

What changes:

  • Native table structures can map more directly to Markdown tables
  • Native list structures can be copied as Markdown lists
  • If document recognition fails, ClariRec falls back to the existing Vision text path and grid-based table detection
  • Older macOS versions keep the previous behavior unchanged

If you edit the recognized text heavily in the result window, structure-based table/list copy actions that depend on the original geometry may hide — because the edited text no longer matches the detected blocks. Re-capture if you need the structured copy again.

Table vs plain text vs formula

Goal Use
Spreadsheet-like grid → Markdown Table copy / .md export
Paragraphs and UI strings Normal OCR copy
Printed math → LaTeX Screenshot to LaTeX (Pro)

Privacy

Default System OCR and structure detection run on your Mac. No table content is sent to ClariRec-operated servers. BYOK OCR is optional and separate — see Local OCR for macOS.

Troubleshooting

No Table button appears
The layout may not look like a regular grid. Try a tighter crop, a clearer image, or paste as plain text and format manually.

Columns look shifted
Merged cells and irregular spacing confuse detectors. Prefer a cleaner source, or fix a few pipes after paste.

Lists instead of tables (macOS 26)
Use the list copy action when the content is a list, not a grid. Document recognition exposes both when available.

You edited the text and the table action disappeared
Expected when structure no longer matches. Capture again for a structured copy, or keep editing as plain Markdown.

Screenshot OCR · Offline · LaTeX · Shortcuts

Download on the App Store