Currently available as Version 0.9 (BETA)

A Django service that automatically syncs and manages files between your local filesystem and AnythingLLM. Just add files to your monitored folder and the system handles workspace organization and embedding automatically.

• File Monitoring 📂: Watches for new, modified and deleted files
• AnythingLLM Integration 🔗:
  • 📤 Automatic file upload and updates
  • 🏢 Workspace management
  • ❌ Cleanup of deleted files
• Image Processing 🖼️:
  • 📷 AI-powered image descriptions using Ollama
  • 🔍 Text descriptions for better search
• Scheduling ⏳:
  • ⏰ Configurable CRON schedule
  • 📧 Manual updates via API endpoint and frontend

• 🐳 Docker and Docker Compose
• 🧠 AnythingLLM instance (works with desktop and docker version)
• 🔑 Access to AnythingLLM Developer API
• 👁️ Optional: Ollama for image description functionality

Quick config via .env file:

ANYTHING_LLM_API=your_api_key        # AnythingLLM API key
ANYTHING_LLM_URL=your_anything_llm_url  # URL without trailing /


USE_CRON=true                        # Enable scheduled checking
CHECK_FILES_CRON=*/1 * * * *         # Check every minute

# Image description (optional)
OLLAMA_URL=http://localhost:11434/api/generate
IMAGE_DESCRIPTION_ACTIVATE=true
IMAGE_DESCRIPTION_MODEL=gemma3:4b
IMAGE_DESCRIPTION_LANGUAGE=english

SORT_FILES=true                      # Auto-organize files
DELETE_UNUSED_FOLDERS=false          # Clean up empty folders

🔑 API Key: AnythingLLM Settings → Tools → Developer API → Generate New API Key

In docker-compose.yml, set your monitored directory:

volumes:
  - C:\MyFolder:/app/AnythingLLM

You can add as many monitored directories you need. If you want multiple, you can do it like this:

volumes:
  - C:\YOUR_FOLDER:/app/AnythingLLM
  - C:\YOUR_SECOND_FOLDER:/app/AnythingLLM_SecondInstance
  - C:\YOUR_THIRD_FOLDER:/app/AnythingLLM3

The programm checks for all folders which starts with "AnythingLLM" within the /app path. Meaning "AnythingLLM_Second", "AnythingLLMHomeWork" and such all get detected, but just putting a number behind it is probably the best solution.

Make sure the path to your folder is before the :/app/AnythingLLM 📂 Each subfolder creates a matching workspace in AnythingLLM:

• C:\MyFolder\Work → Work workspace
• C:\MyFolder\Personal → Personal workspace

Workspaces are automatically created and deleted to match your folder structure.

⚠️ Files within the source folder (in this example the "C:\MyFolder") are skipped. Make sure you create subfolders, else the program can't create workspaces or upload files, as they are not detected.

Generates searchable text descriptions for images:

• 🔍 Detects images (.jpg, .png, etc.) in monitored folders
• 🧠 Uses Ollama for AI description
• 📄 Creates .image_description files
• 📤 Only uploads text descriptions for efficient embedding

Setup: Enable with IMAGE_DESCRIPTION_ACTIVATE=true and configure Ollama settings in .env

⚠️ BETA Feature: Use with caution.

Automatically organizes documents into folders based on workspace associations:

• 🔍 Moves files to folders matching their workspace names
• 📁 Creates workspace folders as needed
• 🔄 Handles single and multi-workspace documents

Configuration:

• Enable with SORT_FILES=true in .env
• Optional: DELETE_UNUSED_FOLDERS=true to clean up empty folders

Note: Only needed if uploading through AnythingLLM Frontend, as this app already manages files automatically.

1. Clone the repository:

git clone https://github.com/MrMarans/AnythingLLM_File-Manager.git

2. Configure your environment:

   • 🛠️ Copy the .env.example file, create an .env file and set your settings.
   • 🔑 Set your API key, AnythingLLM URL, watched folder path
   • 📂 Configure docker-compose.yml with the correct volume mapping for your monitored directory

3. Start the service:

docker-compose up -d

🆙 Updating to a new version or updating the configuration? Use:

docker-compose down
docker-compose up -d --build

1. File Monitoring 🔍:

   • Periodically checks the monitored directory based on the configured CRON schedule ⏳
   • Detects new, modified, and deleted files ✅
   • Maintains a database of detected files 🗃️

2. AnythingLLM Integration 🔗:

   • 📤 New files are automatically uploaded
   • 📝 Modified files trigger updates
   • ❌ Deleted files are removed
   • 🏢 Workspaces are created based on folder structure

3. Image Processing (when enabled) 🖼️:

   • 🔍 Detects image files in monitored directories
   • 🧠 Sends them to Ollama for description generation
   • 📝 Saves descriptions as separate files
   • 📤 Uploads only the text descriptions to AnythingLLM

4. Workspace Management 🏗️:

   • 📁 Creates workspaces automatically for new folders
   • 🔄 Updates embeddings when files change
   • 🧹 Removes empty workspaces to maintain cleanliness

You can modify the CHECK_FILES_CRON environment variable to adjust the checking frequency:

• */1 * * * * - 🔄 Every minute (default)
• 0 */2 * * * - ⏲️ Every 2 hours
• 0 9-17 * * 1-5 - ⏰ Every hour between 9 AM and 5 PM, Monday to Friday

To deactivate CRON Scheduler, set USE_CRON=false in the .env file

You can let the files manually update with a post request to the ip:port/update_files/update/ endpoint. For most people it will be http://localhost:8000/update_files/update/ Also available are:

• update_files/sort/ for sorting files in folders
• update_files/clean/ for deleting empty folders in AnythingLLM
• update_files/scan/ to just check how many file updates there are
• update_files/create_image_descriptions/

The application includes a simple web UI for managing your AnythingLLM files directly from your browser 🌐:

1. How to access: 🔗

   • Open your browser and navigate to the application's base URL (e.g., http://localhost:8000/)
   • You'll see a user-friendly dashboard with action buttons

2. Available actions: 🎮

   • 📤 Full Upload and Cleaning - Upload files and perform cleaning operations
   • 🗂️ Sort Files - Sort files into appropriate folders based on workspace associations
   • 🧹 Clean Folders - Delete unused folders
   • 🔍 Scan Files - Check for new, changed, or deleted files (doesn't upload anything)
   • 🖼️ Create Image Descriptions - Generate descriptions for all images

3. Usage: 👆

   • Simply click on the desired action button
   • The system will execute the requested operation
   • Results will be displayed in the response area below the buttons

This interface provides a convenient alternative to API calls or waiting for scheduled CRON jobs to run. 💫

Common issues and their solutions:

1. Files not being detected 🧐:

   • 🔍 Check the volume mounting in docker-compose.yml
   • 🔑 Verify file permissions
   • 📝 Check the logs for any errors

2. AnythingLLM connection issues 🔌:

   • ✅ Verify ANYTHING_LLM_URL is correct
   • 🔑 Ensure ANYTHING_LLM_API key is valid
   • 🌍 Check network connectivity

3. Schedule not running ⏳:

   • ✅ Verify CHECK_FILES_CRON format
   • 📜 Check container logs for scheduling errors
   • 🌍 Ensure the container has the correct timezone settings

4. Image description not working 🖼️:

   • ✅ Check Ollama is running and accessible
   • 🔍 Verify OLLAMA_URL is correct
   • 📝 Ensure the specified model is available in your Ollama instance

The service logs all operations and errors. Access the logs using:

docker-compose logs -f

For issues, questions, or contributions:

• 📝 Create an issue in the repository
• Make VERBOSE true in the .env file, this will print more logs
• 🛠️ Include logs and configuration details when reporting issues

• 📝 API documentation improvements

• 👤 Face Recognition Feature for Image Description

• 🎯 Check if filetype is supported by AnythingLLM

• 🚀 Much More exciting features!

• Version 1.0 will release with AnythingLLM Desktop Support, tested and working on Linux and Windows and granular API endpoints. No releasedate or window yet clear!

⚠️ No security testing has been conducted. Use at your own risk. This has not been tested for production. Please review the settings.py file and adjust as needed. You have been warned.

🔐 If you have security expertise, pull requests with security improvements are welcome!

MIT