Design a Game Loop
Define a game loop that makes one session feel complete.
A game idea becomes easier to design when the fantasy, core verb, reward rhythm, and session end are named. This Cookbook helps you describe what the player wants to feel, what they do again and again, why the loop rewards them, and how one session closes. Enter only concept facts and feel references. Every prompt tells the assistant not to copy features or invent engine, platform, budget, or technology claims. You leave with a compact loop sheet for a small playable idea.
- Made for
- Game designers, writers, students, and hobbyists shaping a small game idea before feature lists
- You leave with
- player fantasy, core verb, core loop, reward rhythm, session map, fail states, onboarding beat, and loop sheet
- Time
- about 50 minutes
- Steps
- 8 guided steps
- Requires
- any AI you already use (a free chat account is usually enough); sample answers included where available
- Completed runs
- 117 (demo metric)
- Average rating
- 4.6 / 5 (demo metric)
Also in Start Here Selected by SousMeow Recently Added Deep Workflows
Have an account? Sign in.
Stages
3 stages in sequence-
14 steps
Fantasy
Name the fantasy, action, loop, and reward feeling.
-
22 steps
Loop
Map one session and the ways the loop can fail.
-
32 steps
Session
Teach the first beat and pack a loop sheet.
Workflow steps
8 steps · completed in orderStage 1: Fantasy
-
1
Lock player fantasy
State what the player should feel like they are doing and what is out of bounds.
A loop feels scattered when the fantasy is only a theme. This step names the role, promise, and boundaries before mechanics multiply. Common mistakes: copying reference features, adding technology claims, or describing the developer fantasy instead of the player fantasy. Need help if the AI gets broad? Paste again and require one player role. You are ready when the fantasy can guide yes/no decisions.
- Fantasy is player-centered The statement describes what the player feels or does.
- References are not copied Reference games shape feel only, not borrowed features.
- No tech claims appear The output does not assume engine, platform, or production details.
-
2
Name core verb
Turn the fantasy into one repeatable player action.
The core verb keeps the loop from becoming a pile of features. This step names the repeated action and tests whether it supports the fantasy. Common mistakes: choosing a noun, adding several verbs, or sneaking in a bigger genre. Need help if the verb is fuzzy? Paste again and ask for one plain verb. You are ready when most future mechanics can be judged by that action.
- One verb leads The core action is not a list of competing verbs.
- Verb serves fantasy The meaning links back to the approved player fantasy.
- Non-core actions are excluded The loop has clear boundaries.
-
3
List core loop steps
Map the repeated sequence from noticing to action, feedback, reward, and return.
A loop needs a reason to repeat. This step shows what the player observes, does, receives, and wants next. Common mistakes: writing a story summary, skipping feedback, or making rewards unrelated to the verb. Need help if the loop is too long? Paste again and ask for six or fewer repeated beats. You are ready when the last beat naturally points back to the first.
- Loop can repeat The final beat creates a reason to return to the first.
- Choices are real The player has decisions, not only instructions.
- Fragile spots are named Likely weak points are visible before adding features.
-
4
Name reward rhythm
Define when rewards arrive, how feedback feels, and how progression supports the loop.
Rewards should reinforce the core action instead of distracting from it. This step names small feedback moments and a progression tie. Common mistakes: rewarding random taps, adding currencies with no choices, or promising content scale not in the facts. Need help if rewards feel generic? Paste again and tie each reward to the core verb. You are ready when feedback, reward, and progression point in the same direction.
- Rewards follow action Reward timing reinforces the approved loop.
- Feedback fits the feel Moment feedback supports the reference feel without copying features.
- Progression stays on-verb Unlocks create more of the core action, not a separate game.
Stage 2: Loop
-
5
Map one session
Shape a single play session with a beginning, middle, ending, and return hook.
Loops live inside sessions. This step makes one sitting feel complete instead of endless or abruptly cut off. Common mistakes: ignoring session length, ending only when content runs out, or adding calendar and platform claims. Need help if the session sprawls? Paste again and ask for one satisfying sitting. You are ready when the player knows why the session ended.
- Session has an ending The player can tell when one sitting is complete.
- Return hook follows the loop The next-session reason comes from the approved loop or rewards.
- Length is respected The session map fits the stated session length.
-
6
List fail states
Name ways the loop can break and gentle safeguards that keep the session readable.
A loop can fail even when its idea is appealing. This step names confusion, randomness, missing feedback, and unclear endings before more features are added. Common mistakes: treating fail states as player blame, inventing punishments, or assuming technical solutions. Need help if the AI gets harsh? Paste again and ask for soft recovery. You are ready when each failure has a design safeguard.
- Failures are about design clarity The fail states do not blame the player.
- Recovery fits the feel Recovery actions respect the approved fantasy and constraints.
- No tech assumptions slipped in The output avoids engine, platform, and production claims.
Stage 3: Session
-
7
Write onboarding beat
Create the first minute that teaches the loop through one small success.
The first minute should let the player feel the loop, not read a feature list. This step teaches the core verb with one tiny win and postpones everything else. Common mistakes: explaining every system, copying a tutorial pattern, or introducing progression too early. Need help if the beat is crowded? Paste again and ask for one verb, one response, one reward. You are ready when the player can learn by doing.
- Onboarding starts with the fantasy The first minute quickly shows what kind of experience this is.
- Core verb is taught by action The player learns by doing one approved verb.
- Complexity waits Later systems do not crowd the first beat.
-
8
Pack loop sheet
Assemble the fantasy, verb, loop, rewards, session shape, and open questions.
A compact loop sheet keeps the idea testable. This step gathers the approved decisions without turning them into a production plan. Common mistakes: adding features at the end, claiming engine feasibility, or losing the session end. Need help if the sheet grows? Paste again and ask for one page only. You are ready when the next design conversation can start from the same loop.
- Sheet preserves the core loop Fantasy, verb, and loop are all visible together.
- Rewards support progression Reward and progression do not become a separate game.
- Questions stay design-focused Open questions do not invent technology or production claims.
Your information
7 details, answered once- Game name Working title is fine.
- Player fantasy What the player wants to feel like they are doing or becoming.
- Core verb One main action the player repeats. Use a verb.
- Session length How long one satisfying play session should take.
- Progression idea How the player sees change over time. Keep it lightweight.
- Known constraints Tone, scope, audience, time, complexity, or content limits. Do not list engine claims unless they are real facts.
- Reference feel Games you like the feel of, not features to copy.
Finished files
What you export when every step is approved
One Markdown file per approved step, plus kit.html (opens in any browser, offline)
and a README of what you approved.
kit.htmlOffline HTML reader for every approved step01-lock-player-fantasy.mdLock player fantasy02-name-core-verb.mdName core verb03-list-core-loop-steps.mdList core loop steps04-name-reward-rhythm.mdName reward rhythm05-map-one-session.mdMap one session06-list-fail-states.mdList fail states07-write-onboarding-beat.mdWrite onboarding beat08-pack-loop-sheet.mdPack loop sheetREADME.mdManifest of what you approved