A guide to building long-term compounding knowledge infrastructure. See details on GitHub .

LlamaFS

A self-organizing file manager that summarizes file contents with Llama3 and automatically renames and organizes files; supports batch and watch (daemon) modes.

LlamaFS is a self-organizing file manager that uses Llama3 (via Groq or Ollama) to summarize file contents and automatically rename and organize files. It supports batch mode (generate a suggested structure for a directory) and watch/daemon mode (monitor a directory and learn from recent manual renames to proactively organize files). It handles many file types including images and audio.

Key Features

  • Self-organizing organization: automatically suggests and performs file renames and archival based on content and conventions (e.g., timestamps).
  • Live watch daemon: intercepts filesystem operations and learns renaming patterns from user actions in watch mode.
  • Privacy-friendly incognito mode: route requests to local Ollama instead of cloud inference to keep files private.
  • Multimodal support: integrates Moondream, Whisper, etc., for image and audio summarization and annotation.

Use Cases

  • Clean up Downloads and Desktop folders: automatically tidy everyday clutter.
  • Media archival: categorize images, audio, and video by content or date.
  • Automated data prep: organize files for downstream processing and annotation.

Technical Highlights

  • Python backend + Electron frontend, using Groq / Ollama LLM inference for content understanding.
  • High-performance caching for fast suggestions (many operations return <500ms by LLM standards).
  • Configurable FastAPI endpoints for batch processing and integrations.

Comments

LlamaFS
Resource Info
Author iyaja
Added Date 2025-09-15
Tags
OSS AI Agent Utility