← All work 05 · Realtime web product · Solo

Werewolf

A realtime web werewolf game, built solo from scratch.

Solo build · realtime state

Role Project Manager · Developer (solo) Team Solo Period September 2026
View live ↗
Next.js React Supabase Tailwind
Werewolf — Realtime web product · Solo
Home — join or create a room
Home — join or create a room
Waiting lobby before the game starts
Waiting lobby before the game starts
Day phase — village deliberation
Day phase — village deliberation
Voting phase at the village court
Voting phase at the village court
Hunter revenge phase
Hunter revenge phase
End-of-game recap
End-of-game recap
50 Users
1 Person building it
6 Game phases
Role
Solo
Team
1 person
Duration
1 month
Status
Live & in use

Context & problem

Werewolf usually needs a moderator to manage turns and count votes by hand — tedious and error-prone. I wanted a version playable without a moderator, with rules enforced by the system.

Role & responsibilities

I built everything: game rules, phase flow, interface, realtime state in Supabase, and deployment. With no team, every decision — including which features not to build — was mine to make and own.

Planning

Version one could only contain what makes one round playable: room, role assignment, night phase, day phase, voting, and game over. Everything else went to a waiting list. That boundary is what made it ship.

Coordination

Without a team, coordination meant with myself and with users. I tested early and noted where players got confused: when to wait, what was happening, who had voted. Those notes turned straight into interface fixes.

Progress monitoring

The indicator was simple: can a group finish a round without my help. Every "what do I do now?" counted as a defect. After several rounds those questions faded.

Risks & responses

Risk Scope creep from constantly adding feature ideas.
Response First-version boundary written down; other ideas moved to a waiting list.
Risk Players confused by realtime state changes.
Response Phase status shown explicitly and every change given a visual cue.
Risk State bugs stalling the game mid-round.
Response Tested with several real groups before wider announcement.

Outcomes

  • The game runs without a moderator, with rules enforced by the system.
  • Used by around 50 users.
  • Evidence that I can manage scope and ship a product alone.

What I would do differently

Working alone made me realise the value of a second person. With no one to challenge assumptions, some decisions only felt wrong after users tried them. In a team, I want to be the one who asks for that challenge earlier.

Artifacts

Documentation reconstructed from project archives for portfolio purposes.