folder2md4llms¶

Convert a folder and its contents into a single Markdown file for use as LLM context. It walks a directory, formats each file, converts common documents to text, describes binary files, and can condense code to fit a token or character budget.

Quick start¶

# Install (macOS)
brew install henriqueslab/formulas/folder2md4llms

# Or any platform
pipx install folder2md4llms

# Convert the current directory to output.md
folder2md .

# Fit the output to a token budget (enables smart condensing)
folder2md . --limit 80000t

The package is folder2md4llms; the command is folder2md.

What it does¶

• Single-file output. One Markdown file with a directory tree, optional stats, and the contents of each processed file.
• Smart condensing. With --limit, code is condensed to signatures and docstrings to fit a token or character budget instead of being truncated.
• Document conversion. PDF, DOCX, XLSX, PPTX, and notebooks are converted to text (convert_docs, on by default).
• Binary descriptions. Images, archives, and other binaries get a short description rather than raw bytes (describe_binaries, on by default).
• Filtering. Respects .gitignore and a .folder2md_ignore file you can scaffold with --init-ignore.
• Parallel processing. Files are processed concurrently (max_workers, default 4).

Where to go next¶

• Installation
• Your first conversion
• Quick reference
• Configuration
• CLI reference

Citation¶

@software{folder2md4llms,
  title  = {folder2md4llms: convert folder structures into LLM-friendly Markdown},
  author = {Henriques Lab},
  url    = {https://github.com/HenriquesLab/folder2md4llms},
  year   = {2025}
}