← Open Source
oil-oil

wolfcha

AI-powered Werewolf (Mafia) social deduction game where every player is controlled by top LLMs like DeepSeek, Qwen, Gemini, and more

ApplicationsOtherTypeScript
Open on GitHub
Momentum
+0stars in 24 hours0.0%
733
Stars
66
Forks
+6
This week
6
Contributors
Created 2026-01-07 · Updated 2026-10-10 · #13970 today
Top developers
README

简体中文

Wolfcha — Play Werewolf solo

You take one seat. AI players fill the rest.

An AI-native Werewolf game for deduction, bluffing, and chaos on demand.

Play online · Run locally · 游戏规则(中文) · 中文说明

One human. A table that talks back.

Wolfcha recreates the part of Werewolf that is hardest to schedule: a complete table of distinct players. Choose your role, enter an 8–12 seat game, and let the AI handle every other personality, secret, accusation, and vote.

Characters first Table-aware memory Decisions with intent
Each AI has a stable personality layered over a hidden game role. Players follow speeches, votes, deaths, and changing suspicions. They accuse, defend, bluff, follow, or hold back according to their faction goal.

What happens at the table

  1. Night falls — Werewolves choose a target while special roles act on private information.
  2. The table speaks — Every surviving player explains, suspects, misdirects, or pushes a read.
  3. Everyone votes — The group turns conversation into a decision.
  4. The story changes — New deaths and revealed information reshape the next round.

You can play as Villager, Werewolf, White Wolf King, Seer, Witch, Hunter, Guard, or Idiot. Conversations are generated in real time, so the same setup can produce a very different table.

Built for atmosphere

  • Retro visual direction with day/night eye-blink transitions.
  • Lip-sync animation while characters speak.
  • Dedicated role artwork for night actions.
  • Optional AI voice playback and spectator mode.

Project origin

Wolfcha was created at the Watcha × ModelScope Global Hackathon. The name combines Wolf with Cha (猹) — part Werewolf, part spectator watching a table of AI personalities collide.

Local development

Requirements: Node.js and pnpm.

git clone https://github.com/oil-oil/wolfcha.git
cd wolfcha
pnpm install
cp .env.example .env.local
pnpm dev

Open http://localhost:3000. Configure the providers you need in .env.local; the available variables are documented in .env.example.

Tech stack

Next.js 16 · TypeScript · Tailwind CSS 4 · Jotai · Radix UI · Framer Motion · Tiptap

Sponsors

TokenDance

  • TokenDance — core game flow, roleplay, and summaries
  • DashScope — AI capability support
  • Watcha — AI capability and showcase platform support

Roadmap

  • Better mobile play
  • Post-game review and free chat
  • Richer memory, bluffing, and table behavior
  • Special mechanics such as time rewind and AI insight
  • Multiplayer with friends and AI players
  • Community ratings for standout AI personalities

License

MIT