MarkItDown - Qomicex Docs - QML Launcher

MarkItDown

MarkItDown (top.qomicex.markitdown) converts local documents into Markdown text, so other plugins (such as the AI assistant) can parse binary content in PDF / Word / Excel / PPT / images / audio.

Capabilities

Format	Notes
PDF	Local text extraction (pdf.js on the main thread); with a vision model configured, pages are rendered to images and run through structured OCR (tables preserved)
DOCX / XLSX / PPTX	Parsed with mammoth / SheetJS / jszip
HTML / CSV / JSON / XML	Native parsing
ZIP / EPUB	Recursively unpacked, with the files inside converted
Images / audio	Requires an OpenAI-compatible vision / audio model (OCR / transcription)

Registered methods

const res = await __PLUGIN_API__.callPlugin('top.qomicex.markitdown', 'convertFile', {
  base64: '...',            // file content, base64 (use readBytes to get contentBase64 for binary files)
  fileName: 'a.pdf',        // file name (used to detect the format)
  mime: 'application/pdf'   // optional
})
// => { title, markdown, text_content, format } or { error }

getConfig(): read the plugin’s LLM settings (baseUrl / apiKey / visionModel / audioModel / prompt)
setConfig(patch): write LLM settings (used by frontend settings pages through callPlugin)

Dependencies and configuration

Pure document conversion (PDF / DOCX / XLSX / PPTX and so on) needs no configuration
Image OCR and audio transcription need an OpenAI-compatible API: fill in the vision / audio model and key in the “MarkItDown Document Conversion (Document conversion)” section of the AI assistant settings page

Optional add-on

MarkItDown is an optional add-on for the AI assistant: the AI assistant works without it, and its convertDocument tool prompts you to install it. Images and audio return a configuration prompt when no API is configured.