MarkItDown

Convert content from documents, web pages and tables into readable Markdown.

Available in app
Document layouts gather into a clean structured Markdown page.

Overview

Documents into clean Markdown, along the right route. Single Word, HTML, RTF and text files use the app's native conversion, reading structure directly from the file itself; extractable PDFs become Markdown with page markers; scans route to text recognition, and encrypted PDFs unlock first. Batch folders, spreadsheet tables and slide text plus speaker notes run through the bundled tooling — slides in true presentation order, formula cells with no displayable value warned loudly, macros never executed.

Each conversion leaves a traceable record: which route ran, inputs matched one-to-one with outputs, sources kept intact; unsupported formats get routing advice, never a guessed conversion, and nothing goes online. The resulting text is treated as outside content — later steps never execute it or trust its self-declared 'success'.

Text structure is preserved, pixel layout is not promised: figures, styling and complex tables get checked through in-app previews.

Where it helps

  • Batch-convert a folder of spreadsheet and slide reports into searchable Markdown
  • Extract page-marked text from PDF papers for downstream summarizing
  • Keep a re-checkable per-file record of which route ran and what it produced
  • Find out whether a format converts directly, needs recognition first, or isn't supported
#conversion#DOCX#xlsx#pptx#PDF

Examples

Examples describe the request and the shape of the expected deliverable; they are not completed run results.

Example prompt

Recursively convert spreadsheets, slides, CSVs and PDFs under documents/ into Markdown, keep the folder structure, skip shortcuts, and list legacy documents that can't convert.

Expected output

A conversion record listing each file's outcome and counts, the Markdown outputs, and proof the originals were left untouched.

Example prompt

The converted PDF's table columns interleaved wrongly — explain whether that's a text-order artifact or real loss, and tell me how to verify structure and extract text reliably.

Expected output

A diff explanation against the source with page-marked text, plus advice to verify layout-critical tables through in-app previews.

Requirements and limits

Install platforms
iOS: iOS · Android: Android
Script runtime
Runs scripts in the task environment
Minimum app version
1.7.60
Reviewed permission scope

iOS · Android

Capabilities
Remote executionRead workspace filesWrite workspace files
Reviewed tools
Shell commands
Reviewed dependencies
openpyxl==3.1.5 · pdfminer.six==20221105
Review decision
Approved
Review stage
Guest execution qualification
Reviewed on
2026-10-27
Implementation and runtime details
Adaptation stage
frozen

iOS

Tools
Shell commands
Capabilities
Remote executionRead workspace filesWrite workspace files
Runtime environment
linux / riscv64 · guest-verified

Android

Tools
Shell commands
Capabilities
Remote executionRead workspace filesWrite workspace files
Runtime environment
linux / riscv64 · guest-verified

License and source

Source and trace details

License

MIT

MIT (root LICENSE.md)

MIT at the pinned upstream revision; full license and per-file adaptation/asset audit are preserved in references/licenses/. Floe additions and synthetic fixtures are identified separately.

Image and material attribution

Image attribution
Floe original, AI-assisted illustration.
Source
https://github.com/K-Dense-AI/scientific-agent-skills
Upstream path
skills/markitdown
Upstream pin
593377bb04
Source checked
2026-10-26
Publisher
JiangNanGenius
Publication commit (retrieval)
25e07dbeeb6cc93c2fe83ddc72a34302cc40832a
Audited source commit
84c524b6fb483e4cad94db076e7d9c9962966c19
Signed catalog digest
iOS 46196ba57ed7ea7f22085e74b113207f31193e820ab8983cf7a60eee953f2340 · Android 16de89d019e2057804a40f215485ccb09d5495a25864b34d681168eeaf8f9697
Publication snapshot (UTC)
2026-10-31

Publication commit 25e07dbeeb, audited source 84c524b6fb; catalog digest iOS 46196ba57ed7… · Android 16de89d019e2….

Versions and status

Version records ship with the signed catalog; the target and released versions are shown below.

Adaptation
Ready
Publication state
Curated skills are available
Deep adaptation
Yes
Released version
1.1.0