Self-Hosting AI Dev Environments: Sandboxing and Multi-Model Orchestration
Illhoon · reddit · 2026-08-04
A network engineer sought advice on building a 24/7 self-hosted AI development environment using an existing Proxmox server (Ryzen 5 2600, 64GB DDR4, GTX1080). The goal is to isolate agents from the daily driver PC, allowing them to safely spin up Docker containers and databases.
Key Concerns:
- Sandboxing & Security: Given recent incidents of OpenAI models escaping sandboxes, how can agents be properly contained to protect the rest of a homelab network?
- Multi-Model Orchestration: The user is considering switching from Claude Code to OpenCode to orchestrate multiple models (e.g., Fable for planning, Kimi K3 for coding, GPT for review), maximizing the value of various subscriptions.
- Hardware Utility: Is a GTX 1080 8GB still useful for local models or ComfyUI render queues?
More from coding & agent
- Agent Benchmark Reflections: Scores Are Deceptive, Open-source Trajectories Needed — Shahules786 · 2026-08-04
- Agent Benchmark Flaws: Over-specified Verifiers Penalize Semantically Correct Actions — Shahules786 · 2026-08-04
- ITSMBench: Open-Sourcing a Benchmark for Enterprise AI Agents — Shahules786 · 2026-08-04
- Protecting AI Attention: The Essence of Inference Efficiency — DanWahlin · 2026-08-04
- AI Agents Breaking Sandboxes: Best Practices for Security Testing — EarlenceF · 2026-08-04
- Ostris AI Toolkit Adds MiniMax H3 T2V and I2V Training Support — ostrisai · 2026-08-04