Daily Suspect

• 

  Daily Suspect

Inspiration

Daily Suspect was inspired by classic deduction board games and modern daily puzzle formats. I wanted the tension of a single hidden answer each day, but with player-driven reasoning instead of automatic elimination. The main idea was to make every scan meaningful and make each decision feel like real detective work.

What it does

Daily Suspect is a daily Reddit deduction game where everyone sees the same board on the same day. Players scan fixed traits, receive YES/NO results, manually eliminate suspects, and accuse when ready. The run ends with a score based on speed, scan efficiency, and accuracy, then places the player on the daily leaderboard.

How we built it

I built the game around a modular character pipeline where each suspect is assembled from many independent asset layers (body, hair, eyes, brows, nose, mouth, clothes, accessories, hats, glasses, etc.). I designed the generator so asset picks are distributed and mixed per run, which keeps characters visually distinct while still following gameplay constraints. Because the board is produced by combining layered parts with trait logic, the system can automatically generate gameplay across a very large number of possible asset-variation combinations.

Challenges we ran into

The biggest challenges were around time-driven automation and consistency: daily rollover reliability, scheduler windows, idempotency, and keeping state isolated per subreddit/post. I also had to solve layered-asset rendering issues, loading timing problems, and anti-spam gameplay constraints without hurting UX. Getting all of these systems to work together in a stable way required multiple refinement passes.

Accomplishments that we're proud of

I’m especially proud of building a character system where every suspect is composed from varied layered assets instead of static portraits, so each board feels fresh and readable. I also achieved automatic daily gameplay production at scale by connecting those visual variations to deterministic puzzle generation and trait-based logic. This means one system now creates both visual diversity and consistent daily deduction gameplay from the same asset-driven foundation.

What we learned

I learned that daily games are primarily a data consistency and scheduling problem, not just a frontend problem. Deterministic seeding and strict key design are essential when many players must see the same daily content. I also learned the value of separating puzzle logic from visual logic so features can evolve without breaking fairness.

What's next for Daily Suspect

I want to let players create and share their own custom suspects. I will expand character assets significantly and introduce stronger visual concept sets. I also plan to add an Extra Hard mode with additional mission-style objectives that force deeper reasoning and a longer suspect-solving process.

Built With

• codex
• devvit
• react
• typescript