pyragify Skill

Source

• Repo: https://github.com/ThomasBury/pyragify
• PyPI: pip install pyragify
• Version: v0.2.0 (April 2026)
• License: Unlicense (fully free)

What It Does

Converts any code repository or document directory into clean, semantically-chunked .txt files optimized for NotebookLM ingestion. Run this BEFORE uploading to NotebookLM.

Setup

pip install pyragify

Usage

CRITICAL: Always use relative portable paths to execute the OS Engine so the workspace can be ported to new machines without breaking:

# With a config file (recommended)
.\.venv\Scripts\pyragify.exe --config-file config.yaml

# Or inline
.\.venv\Scripts\pyragify.exe \
  --repo-path /path/to/repository \
  --output-dir /path/to/output \
  --max-words 200000 \
  --skip-dirs "__pycache__" \
  --skip-dirs "node_modules" \
  --verbose

Example config.yaml

repo_path: c:\Users\BAB AL SAFA\Desktop\open-workspace
output_dir: c:\Users\BAB AL SAFA\Desktop\open-workspace\.core\toolbox_library\core.toolbox\notebooklm\skills\pyragify\output
max_words: 200000
max_file_size: 10485760  # 10 MB
skip_patterns:
  - "*.log"
  - "*.tmp"
skip_dirs:
  - "__pycache__"
  - "node_modules"
  - ".git"
verbose: false

Output Structure

output/
  python/      ← Python functions, classes, comments
  markdown/    ← Markdown sections split by headers
  html/        ← HTML script and style chunks
  css/         ← CSS rule chunks
  other/       ← All other readable file types
  remaining/   ← Overflow chunks after word limit
  metadata.json
  hashes.json  ← MD5 hashes for incremental re-runs (only reprocesses changed files)

Supported Inputs

• .py — chunked by function/class/comment
• .md, .markdown — split by header sections
• .html — script and style chunks
• .css — CSS rule chunks
• All other UTF-8 readable files — included as plain text

Workflow with notebooklm-py

# Step 1: Preprocess the workspace
pyragify --config-file config.yaml

# Step 2: Upload output chunks to NotebookLM
notebooklm create "Workspace Analysis"
notebooklm use <notebook_id>
notebooklm source add "./output/markdown/chunk_0.txt"
notebooklm source add "./output/python/chunk_0.txt"

# Step 3: Query and generate
notebooklm ask "What are the architectural gaps in this workspace?"