Dev builds Werewolf game where AI agents bluff and vote alongside humans
rezakhan0 · reddit · 2026-11-15
A Reddit developer shared a Werewolf game where multiple AI agents sit at the same table as human players. Each agent holds a secret role, speaks out loud, bluffs, and votes.
How it's built:
- One model call per agent per turn; each agent only sees what its player would legitimately see, plus private notes, so secrets can't leak.
- Claimed roles never change the actual assignments — agents must infer the truth from conversation, just like humans.
The hardest parts, per the author, were making agents bluff convincingly and keeping the system fast and cheap enough for live voice, quick texts, and reasoning.
More from coding & agent
- AI-assisted hackers crack Nikon firmware encryption, flashing custom features onto Z8 and other cameras — jedisct1 · 2026-11-15
- Matt Turck mocks agentic overkill: 12-step workflow to auto-send one email — mattturck · 2026-11-15
- Dev argues sidebar-plus-agent-list UI is insufficient for multi-machine agent orchestration — kevinkern · 2026-11-15
- Screenshot leaks Antigravity model list: Gemini 4 Argon, Hydrogen, and Orchestrator — Smart-Knee-5596 · 2026-11-15
- Dev builds a full Git Tag manager UI with Copilot Canvas in VS Code — mariorod1 · 2026-11-15
- A playable tank game built with Claude Code: Opus plans, parallel Sonnets code in worktrees — IamHuggos · 2026-11-15