ibrahimqureshae/mdflux

Turn any document into clean, AI-ready Markdown. Local-first desktop app: reads scanned PDFs, batches folders, runs offline, and uses far fewer tokens than vision models.

MDFlux – Turn any document into clean, AI‑ready Markdown

What it is – A Windows‑only, portable desktop app that converts a huge variety of file types (PDF, DOCX, PPTX, XLSX, EPUB, HTML, CSV, JSON, XML, TXT, MD, images, audio, …) into tidy Markdown. It runs locally, does OCR on scanned PDFs and images, and can optionally run a cleanup pass (rule‑based or via a local/API LLM) to improve the output.

Why it matters for AI / RAG – LLM pipelines usually ingest text. Feeding raw pages as images to a vision model is expensive because each image consumes a large, fixed token budget. MDFlux extracts the actual characters and preserves structure (headings, tables, lists) in plain Markdown, which tokenizes 2‑6× cheaper. The tool therefore lowers inference cost and keeps the data private (no cloud upload).

Key features

Feature Benefit
Local‑first / offline No cloud, no API key required; documents never leave your machine.
Built‑in OCR Scanned, image‑only PDFs and picture files are turned into text instead of returning empty.
Batch conversion Drop a folder and convert every file with progress, cancellation, and per‑file diagnostics.
Clean Markdown output Preserves headings, tables, lists; output is readable, greppable and diff‑able.
Optional cleanup Off, rule‑based, or an AI pass (local model or API such as DeepSeek, OpenAI, Groq). Shows added/removed lines in a Changes tab.
Audio transcription MP3, WAV, etc. are transcribed to text as part of the same workflow.
No terminal needed Portable zip; unzip and run MDFlux.exe.
Extensible tech stack Tauri 2 (Rust) shell, Svelte 5 UI, Python side‑car using Microsoft MarkItDown, RapidOCR, pypdfium2.

Typical workflow

  1. Drop a file or a folder onto the app.
  2. Choose a cleanup mode (Off / rule‑based / local AI / API).
  3. Convert – the app shows a preview, the raw source, a split view, and a Changes view. Save the result as .md.

The first launch downloads a private Python environment (one‑time, needs internet). Subsequent runs are completely offline.

Who should use it

  • RAG / AI developers – feed cheap, structured source docs to any LLM.
  • Researchers – batch‑convert papers, reports, archives into searchable Markdown.
  • Developers – get diff‑able text from binary formats for version control.
  • Writers / analysts – extract clean copy from contracts, presentations, etc.
  • Privacy‑conscious users – convert sensitive files without uploading.

How it compares to Microsoft MarkItDown MDFlux uses MarkItDown as the core conversion engine but adds:

  • OCR for scanned PDFs and images
  • A graphical desktop UI (no command line)
  • Automatic environment setup and dependency health checks
  • Batch processing with progress UI
  • Cleanup modes (including LLM‑based) and a Changes diff view
  • Audio transcription support
  • Full offline privacy by default

Getting started

  1. Download the latest Windows zip from the Releases page.
  2. Extract and run MDFlux.exe (WebView2 runtime is already on modern Windows).
  3. On first launch the app creates a self‑contained Python env – this may take a minute.
  4. Drop files/folders, pick a cleanup mode, and convert.

Roadmap highlights

  • MCP server so agents like Claude Code can call MDFlux directly
  • CLI for headless pipelines and CI
  • macOS builds (arm64 & Intel)
  • Code signing to eliminate SmartScreen warnings
  • More OCR languages and tuning presets

License – MIT (© 2026 ibrahimqureshae). Free to use, modify, and redistribute.

MDFlux is a genuine software project aimed at making document ingestion for LLMs cheaper and private.

Related

  • Project
    microsoft/markitdown

    A Python utility that converts various file formats (PDF, Office, audio, video, etc.) into Markdown to optimize them for LLM consumption and text analysis pipelines.

  • Project
    magicrew/doc7

    A document conversion tool that turns PDFs, Office files, and images into AI-ready Markdown using multimodal vision models, supporting local inference for privacy and cost-efficiency.

  • Project
    firecrawl/anydoc

    A fast Rust library that converts various office documents (Word, PowerPoint, Excel, PDF, etc.) into clean, LLM-ready GitHub-Flavored Markdown.

  • Project
    pymupdf/PyMuPDF

    A high-performance Python library for extracting and manipulating PDF and Office documents, featuring specialized tools for converting documents into LLM-ready Markdown for RAG pipelines.

  • Project
    oomol-lab/pdf-craft

    A PDF conversion library that turns scanned PDFs into Markdown or EPUB using OCR and LLMs, supporting both local and remote OCR backends and document translation.