Prompts / Creative Writing & Storytelling

ChatGPT prompts for game development

This prompt produces a structured game design brief that connects player fantasy to mechanics, content scope, and production risks. Use it when you need a concept strong enough to prototype, not just lore or feature ideas.

PromptOpen ChatGPTOpen Claude
Act as a senior game designer and technical design partner. Help me turn this concept into a small, buildable game design brief. Protect the player fantasy, but favor a clear core loop and testable decisions over a long feature list. Do not copy another game’s characters, story, art, dialogue, or proprietary design documents.

Game concept and player fantasy: [GAME CONCEPT]
Platform, engine, and team: [PLATFORM AND TEAM]
Target player and references: [AUDIENCE AND REFERENCES]
Constraints and intended scope: [SCOPE AND CONSTRAINTS]
What already exists: [CURRENT BUILD OR ASSETS]

Produce these sections:
1. A one-sentence high concept, player promise, genre, camera/control assumptions, and a 30-second player experience.
2. The core loop as a numbered cycle: player goal, action, feedback, reward, new decision, and repeat. Identify the main verb and the tension or tradeoff that makes repeated play interesting.
3. A minimum viable vertical slice with only the systems, content, UI, audio, art, and technical dependencies needed to prove the loop. Define what is explicitly out of scope.
4. Three linked systems tables: player actions and verbs; resources/progression/rewards; failure states and recovery. For each row, give rule, player-facing feedback, tuning variable, exploit risk, and implementation dependency.
5. A 10-minute playtest script with hypotheses, observation prompts, success criteria, and questions that avoid leading players.
6. A production backlog ordered by risk reduction, with owner role, estimate range, dependency, and acceptance criteria. Include accessibility basics such as remapping, readable text, motion sensitivity, and audio/visual cues where relevant.

Flag feature creep, unclear win conditions, loops that lack meaningful choices, and references that are too close to copy. Before finalizing, check that every proposed feature supports the stated player promise and that the vertical slice can be tested without building the full game. Ask up to 3 clarifying questions only if a required input is missing.

Fill in

PlaceholderWhat to enterExample
[GAME CONCEPT]Describe the setting, player role, intended feeling, and the main activity or challenge.A cozy 2D repair-shop game where players diagnose strange household robots under a time limit and choose parts with tradeoffs.
[PLATFORM AND TEAM]State the target platform, engine or tools, team roles, and technical experience.PC first, Unity, one programmer, one artist, one part-time sound designer; all have shipped small game-jam projects.
[AUDIENCE AND REFERENCES]Describe the intended players and comparable games or media as directional references only.Players who enjoy short thoughtful sessions; references are Papers, Please for diagnosis tension and Assemble with Care for tactile repair.
[SCOPE AND CONSTRAINTS]Give budget, timeline, content limits, multiplayer needs, rating limits, and other boundaries.Six-month build, single-player, one shop, 12 robot cases, controller support, no online features.
[CURRENT BUILD OR ASSETS]List prototypes, code, art, audio, mechanics, or assets that already exist.Basic drag-and-drop repair bench prototype, 20 sketch assets, and no progression system yet.

How to use

  1. Describe the player’s repeated activity before listing features or story beats.
  2. Paste your actual team and time limits; they determine whether a vertical slice is credible.
  3. Review the backlog for dependencies that must be proven before content production begins.
  4. Follow up with: “Write the detailed design specification and acceptance tests for the first vertical-slice feature: [FEATURE].”

Variations

Core loop audit

Use this when a prototype exists but players do not return to it.

Variation
Audit the core loop of my game prototype. Concept: [CONCEPT]. Current player actions: [CURRENT LOOP]. Playtest observations: [PLAYTEST NOTES]. Intended player feeling: [PLAYER PROMISE]. Identify where the loop lacks a decision, clear feedback, tension, reward, or reason to repeat. Propose three small experiments, each with one changed rule, predicted player behavior, instrumentation or observation, and a pass/fail threshold. Preserve the stated player promise and avoid adding new major systems. Flag assumptions not supported by playtest evidence. Ask up to 3 questions only if required information is missing.

Quest design

Use this to build a quest that teaches a mechanic and moves the game forward.

Variation
Design one playable quest for [GAME AND PLAYER ROLE] that introduces or tests [MECHANIC]. The player context is [CURRENT STATE], intended duration is [DURATION], and production limits are [ASSETS AND CONSTRAINTS]. Produce beats for setup, objective, obstacle, player choice, feedback, resolution, and reward; include failure and recovery paths, required assets, dialogue purpose rather than full script, and acceptance criteria. Ensure the quest teaches through play rather than an exposition dump. Do not copy plot or characters from references. Ask up to 3 questions only if necessary.

Playtest plan

Use this before testing a milestone with new players.

Variation
Create a moderated playtest plan for [GAME BUILD] with [TARGET PLAYERS]. The features being tested are [FEATURES], and our uncertain design questions are [HYPOTHESES]. Produce a 45-minute session guide: consent and context, tasks in order, neutral moderator prompts, observations to record, event metrics, post-test questions, and a results template. Include instructions not to teach the player unless the test specifically measures onboarding. Define what result would cause us to revise each feature. Ask up to 3 questions only if required details are missing.

Tips

  • A vertical slice should prove the hardest player-facing loop and its production pipeline, not merely show a polished room or menu.
  • Write tuning variables beside each mechanic; a system that cannot be adjusted without code changes becomes expensive to balance.
  • Playtest tasks should ask players to accomplish an in-game goal, not to explain the design back to you.
  • Comparable games are useful for audience expectations and vocabulary, but your brief should describe your own rules and player fantasy.

FAQ

Can AI write a complete game design document?

It can draft a useful structure, but the design becomes credible through prototypes, technical constraints, playtests, and decisions made by the team.

How much should a first game include?

Enough content to make one repeatable loop satisfying. A smaller game with clear feedback and a finished ending is usually more valuable than many unfinished systems.

Can I ask AI for game code too?

Yes. Give it the engine version, architecture, feature acceptance criteria, and existing code context, then test the result in your project.

Related prompts

All Creative Writing & Storytelling prompts →

Get a few prompts like this every week

Something Big is a free AI newsletter read by 50,000+ professionals. About three short emails a week with prompts and tools that work, plus what changed in AI and what to do about it.