AI Werewolf Kill
A single-player werewolf killing web game driven by the Big Language Model (LLM). The 11 AI players have independent personalities, memory and reasoning abilities, and are able to disguise, deceive and analyze logically. Visual novel-style interface, typewriter effects, immersive reasoning experience.
Single-player werewolf killing web game driven by Big Language Model (LLM) You need to configure your api and key first, and you can use DeepSeek for verification in China. ##Game characters ###Good people camp (8 people) | role| number| skills| |------|------|------| | villagers| 4 |No special skills, rely on reasoning to vote| | Prophet| 1 |Identify one person every night| | witch| 1 |Antidote saves people/poison kills people| | guards| 1 |Protecting one person every night from werewolves| | Hunter| 1 |You can shoot one person away when you are out| ###Werewolf Camp (4 people) | role| number| skills| |------|------|------| | werewolf| 3 |Participate in killing discussions every night| | Wolf King| 1 |You can shoot one person away when you are out| ##Game flow ``` Night stage day phase │ │ 丨 ─ The werewolf discusses killing ── ─ ─── ─ ─ ─ ─ ─ ─ ─► Announced his death last night 丨 ─ The witch chooses to save/poison │ 丨 ─ Prophet Inspection 丨 ─ Speech and discussion ─ Guard 丨 ─ Vote for exile ─ Last words (if any) ``` ##Playing skills ###As a good person - ** Listen carefully to the speech **: Analyze everyone's logical chain - ** Remember key information **: Who tested whom, who protected whom - ** Be wary of standing on the side **: Standing blindly may be used by werewolves ###As a werewolf - ** Unified statement **: Cooperate with teammates to fabricate logic - ** Reasonable disguise **: Imitate the good person speaking model - ** Protect wolf teammates **: Don't expose yourself too early
