Guard Lab: open-source 3D game where an LLM guards a safe and you try to trick it
flowersslop · x · 2026-10-12
Developer flowersslop released guard-lab, an open-source 3D experiment where an LLM plays a safe guard and the player must lie, persuade, or break in to steal the code.
- Play by chatting with the guard, or go loud and see how it reacts
- You can play as the robot guard or run AI vs AI adversarial experiments
- Works via the Codex app, other agent bridges, local models (Ollama), or compatible APIs
- Built on Electron + Three.js for Windows/macOS/Linux, MIT licensed for forking and modding
It's essentially a playful arena for prompt injection and jailbreak-style attacks on agent guardrails.
Related event: Guard Lab: Open-Source Game Tests If You Can Trick an LLM Guard(3 posts)→
More from coding & agent
- Practitioner's verdict on autoresearch: great at speeding experiments, not at frontier runs — iaindunning · 2026-10-12
- levelsio cancels nearly all SaaS subscriptions: money now goes only to AI inference, content, data centers, power and taxes — karlwaldman · 2026-10-12
- Grok Bot is winning the agentic personal assistant race, says dev — Arindam_1729 · 2026-10-12
- Does Codex refuse or cut back work when its own budget estimate runs high? — r618NecessaryStation · 2026-10-12
- Excalidraw for slides is the GOAT: MCP integration plus animations, no AI slop — HamelHusain · 2026-10-12
- Autoresearch for LLM pretraining needs humans in the loop; stacked changes are the killer — menhguin · 2026-10-12