Herd

Document and code management toolkit that uses language models for file organization, content extraction, and document analysis. Successor to llamaherder.

Features

• Smart File Renaming: Automatically generate meaningful filenames based on content
• Code Snippet Extraction: Extract and organize important code segments
• Citation Management: Extract and format citations from research documents
• Document Idealization: Create canonical, cleaned versions of text content
• Image Processing: Generate alt text and description for image files
• Documentation Generation: Create comprehensive documentation for code repositories
• Deduplication: Find and manage duplicate files efficiently
• Interactive CLI: User-friendly interface with intuitive navigation and directory switching

Installation

Install from PyPI:

pip install geepers-herd

The CLI installs as herd and the Python module as herd_ai.

System requirements

The toolkit pulls in heavyweight dependencies (Pillow, reportlab, beautifulsoup4, rich). For local inference it expects an Ollama server; point OLLAMA_API_URL at a non-default host if yours isn't on http://localhost:11434.

Quick Start

# Process files in a directory (CLI)
herd --dir ~/Documents/project --rename --recursive

# Extract code snippets 
herd --dir ~/code/project --snippets

# Generate documentation
herd --dir ~/code/project --docs

# Process images
herd --dir ~/images --images

# Launch interactive CLI
herd

# Use in your Python code
from pathlib import Path
from herd_ai.rename import process_renames
from herd_ai.snippets import process_snippets

# Smart file renaming
process_renames(Path("~/Documents/project"), recursive=True, provider="ollama")

# Extract code snippets
process_snippets(Path("~/code/project"), recursive=True, provider="ollama")

LLM Provider Support

Herd supports multiple LLM providers:

• Ollama (default): Uses local models like Llama, Gemma, Mistral
• xAI: Uses Grok models for advanced processing
• Google Gemini: Uses Google's Gemini API for text and image processing
• OpenAI: Uses OpenAI's API for text and image processing
• Cohere: Uses Cohere's API for text processing

Select your provider with the --provider flag:

herd --dir ~/Documents --rename --provider openai

Or in code:

process_renames(Path("~/Documents"), provider="openai")

Interactive CLI

Herd features an intuitive interactive CLI with the following keyboard shortcuts:

• Numeric keys (1-9): Select menu options
• s: Access settings menu (change provider, set API keys)
• d: Change target directory without restarting
• q: Quit the current menu or application
• Ctrl+C: Go back one menu level (or exit if at main menu)

Launch the interactive CLI:

herd

License

MIT. Author: Luke Steuber (luke@lukesteuber.com, lukesteuber.com).