OffgridDev™

​​​​​​​    #InfiniteTokens

Description

OffgridDev™ is a local, offline-first AI development environment for developers and researchers who want serious AI assistance without the cloud, tokens, or subscriptions.

Built as an open-source, air-gapped stack, OffgridDev™ orchestrates Docker, Ollama, and Continue for VS Code to keep sensitive codebases, research scripts, and algorithms strictly on your local machine. The stack auto-detects your GPU and available VRAM to match open-weight models to your exact hardware limits while blocking choices your system can't run. Out of the box, it ships with curated, VRAM-tiered defaults across coding, reasoning, general-purpose, and vision model families — Qwen (2.5 Coder, Qwen3 Coder, Qwen3), DeepSeek (Coder and R1), Gemma (3 and 4), Devstral, QwQ, Hermes 3, Phi, and Moondream/Qwen2.5-VL — each pulled under a consistent, friendly name shared across Open WebUI, VS Code's Continue extension, and FieldStationAI.

Quick Start Guide

1. Install Docker Desktop and make sure it's running. Note: this may require administrator or super user rights.
2. Download this repo and extract it to C:\OffgridDev (Windows) or ~/offgriddev (macOS/Linux).
3. Optionally edit config/models.env to choose which models to pull — sane defaults are already set.
4. Start the stack:
   • Windows: powershell -ExecutionPolicy Bypass -File scripts\offgriddev-start.ps1
   • macOS / Linux: ./scripts/offgriddev-start.sh
5. Open http://localhost:3000 to chat, or use VS Code's Continue extension (config written automatically). Alternatively, you can download FieldStationAI, launch it, and go to Menu -> Enable Ollama. It will automatically detect the local Ollama API and connect.
6. Stop the stack when done:
   • Windows: powershell -ExecutionPolicy Bypass -File scripts\offgriddev-stop.ps1
   • macOS / Linux: ./scripts/offgriddev-stop.sh

For hardware requirements, full setup steps, and troubleshooting, see Documentation below.

Documentation

• Detailed setup and usage instructions are available at:
  • Hardware & Model Configuration — GPU/VRAM/disk requirements, supported model families, and the config/models.env reference.
  • Installation & Usage — full install steps, starting/stopping the stack, the Open WebUI web interface, and VS Code/Continue integration.
  • Troubleshooting — container logs, tool-calling errors, RAG embeddings, GPU fallback detection, and resetting a stuck container.

Additional Resources

• Docker Desktop
• Ollama
• Open WebUI documentation
• Continue for VS Code
• FieldStationAI

About the Team

The Mobile Technologies Core provides investigators across the University of Michigan the support and guidance needed to utilize mobile technologies and digital mental health measures in their studies. Experienced faculty and staff offer hands-on consultative services to researchers throughout the University – regardless of specialty or research focus.

Learn more at: https://depressioncenter.org/mobiletech.

Contact

To get in touch, contact the individual developers in the check-in history.

If you need assistance identifying a contact person, email the EFDC's Mobile Technologies Core at: efdc-mobiletech@umich.edu.

Credits

Authors:

• Gabriel Mongefranco (@gabrielmongefranco)

Contributors:

• Eisenberg Family Depression Center (@DepressionCenter)

This work is based in part on the following projects, libraries and/or studies:

• Ollama: Runs open-weight language models locally and serves them over a local API. This project's model runtime.
• Open WebUI: Self-hosted chat interface for local LLMs. Provides this project's web UI at localhost:3000.
• Continue: Open-source AI coding assistant for VS Code. Provides in-editor chat, edit, and autocomplete against the local Ollama models.
• Docker / Docker Compose: Containerizes and orchestrates the Ollama and Open WebUI services.

License

Copyright Notice

Copyright © 2026 The Regents of the University of Michigan

Software and Library License Notice

This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.

This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.

You should have received a copy of the GNU General Public License along with this program. If not, see https://www.gnu.org/licenses/gpl-3.0-standalone.html.

Documentation License Notice

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.3 or any later version published by the Free Software Foundation; with no Invariant Sections, no Front-Cover Texts, and no Back-Cover Texts. You should have received a copy of the license included in the section entitled "GNU Free Documentation License". If not, see https://www.gnu.org/licenses/fdl-1.3-standalone.html

Citation

If you find this repository, code or paper useful for your research, please cite it.

Citation Example:

Mongefranco, Gabriel (2026). OffgridDev. University of Michigan. Software. https://github.com/DepressionCenter/OffgridDev
​​​​​​​    DOI: pending

---

Copyright © 2026 The Regents of the University of Michigan