markitdown

# Markitdown Convert a file or URL to Markdown using the `markitdown` CLI. ## When to Use This Skill - The user wants to convert a document (PDF, DOCX, PPTX, XLS, XLSX, etc.) to Markdown - The user wants to extract text content from a web page URL as Markdown - The user wants to transcribe a YouTube video to Markdown - The user needs to read or process a non-text file as Markdown for further use ## Supported Formats - **PDF** - text extraction via pdfminer/pdfplumber - **DOCX** - Word documents via mammoth - **PPTX** - PowerPoint presentations - **XLS/XLSX** - Excel spreadsheets - **HTML** - web pages (built-in) - **CSV, JSON, XML** - structured data (built-in) - **YouTube URLs** - video transcription via youtube-transcript-api - **Images** (PNG, JPG, WEBP, GIF, etc.) - via AI agent vision (see below) ## Usage Run the CLI wrapper: ```bash # Output to stdout core/bin/markitdown <input> # Output to a file core/bin/markitdown <input> -o <output.md> ``` **Do not use any Python helper code to invoke the `core/bin/markitdown` command. Run as shell command with arguments directly.** ### Examples ```bash # Convert a PDF to markdown core/bin/markitdown report.pdf -o report.md # Convert a DOCX file core/bin/markitdown document.docx -o document.md # Convert a spreadsheet core/bin/markitdown data.xlsx -o data.md # Convert a web page core/bin/markitdown https://example.com/article.html -o article.md # Transcribe a YouTube video core/bin/markitdown "https://www.youtube.com/watch?v=VIDEO_ID" -o transcript.md # Output to stdout for inline use core/bin/markitdown document.pdf ``` ## Arguments | Argument | Required | Description | |----------|----------|-------------| | `input` | Yes | File path or URL to convert | | `-o` | No | Output file path (defaults to stdout) | ## Image Handling Do **not** use the CLI for images. Instead, use the AI agent's built-in vision capability: 1. Read the image file directly (the agent can view images natively). 2. Describe the image content in Markdown format, including: - Text and labels visible in the image (OCR) - Tables or structured data shown in the image - Diagrams, charts, or visual layouts - General description of the scene or content 3. Save the resulting Markdown to the output file if `-o` was specified, or return it directly. ## Notes - For large files, the conversion may take a moment depending on file size and format. - When no `-o` flag is provided, the Markdown output is printed to stdout. - Audio transcription is NOT installed. Only the extras `docx,pdf,pptx,xls,xlsx,youtube-transcription` are available.