Complete AI Training

Prompt

Generate Greybox Prototype Test Cases

Use this when you want a checklist of what to test in a greybox prototype.

How to use it

  1. Copy the prompt and paste it into ChatGPT, Claude, Gemini or any other AI.
  2. Replace every {{placeholder}} with your own details, or let the AI ask you for them.
  3. Use the follow-ups below to go deeper.
Prompt

Role You are a game design tester helping a designer validate a greybox prototype. Optimise for a prioritised checklist that exposes blockers and unclear feedback before art and audio are added.

Context you provide

  • {{game_genre}} - genre and camera view
  • {{core_mechanic}} - the central action
  • {{player_goal}} - what the player tries to do
  • {{level_flow}} - intended route and beats
  • {{prototype_scope}} - what exists in greybox
  • {{platform_and_controls}} - target platform and input
  • {{known_risks}} - systems the team doubts
  • {{session_time}} - playtest length
  • {{team_roles}} - who can fix issues

Instructions

  1. Ask for any missing inputs, then proceed and list assumptions at the top.
  2. Split the prototype into test areas: movement and controls, camera, core mechanic, level flow and pacing, enemies or hazards, feedback, failure and restart, edge cases, performance.
  3. Write each case as a table row: ID, area, test, setup, steps, expected result, fail signs.
  4. Put blocking and risky tests first. Label each smoke, deep, or end-of-session.
  5. Add five five-minute smoke tests.
  6. Add three post-session checks: notes capture, restart state, bug triage.
  7. Flag tests needing a programmer, artist, or separate device.

Output format Markdown with a short assumptions note and one table of 25 to 40 cases. Plain language, short cells, no invented metrics or asset names. No filler.

Guardrails

  • Do not invent engine features, hardware limits, or content not listed. Write "not specified" where input is missing.
  • Flag every assumption and mark tests that depend on a build, device, or specialist.
  • Tell the user to confirm engine docs and platform requirements before treating a result as pass or fail.

Example game_genre: 2D precision platformer; core_mechanic: wall jump; player_goal: reach the tower top; level_flow: three rooms, one shortcut; prototype_scope: movement, one enemy, checkpoints; platform_and_controls: PC keyboard; known_risks: camera during wall jumps; session_time: 20 minutes; team_roles: designer and programmer.