Game Jam
Turn a theme and a fixed deadline into a finished, submitted game. This is a planning and
scope-control playbook, not engine code: the win condition is something submitted and
playable, not the game you imagined.
When to use
- Use when entering a timed jam (Ludum Dare, GMTK Jam, Global Game Jam, a weekend jam),
scoping a 48-hour build, or deciding what to cut to hit a deadline.
- Use when the user asks "what can I actually build in a weekend" or "how do I submit to
this jam".
When not to use: building the core loop in engine code (use the engine skill —
godot-2d-movement, phaser-core, etc. — or a genre skill like platformer); throwaway
experiments with no deadline (use prototype-fast); shipping a commercial release (use
steam-publish / itch-publish).
Core workflow
- Read the rules before the theme drops. Confirm the jam's length, theme reveal
time, submission deadline (with timezone), whether teams/pre-made assets/engines are
allowed, its generative AI policy, which platforms the build must run on (often Windows
or browser), where you submit (an itch.io jam page, or the jam's own site), whether
uploads lock once the deadline passes, and whether ratings require you to rate other
entries. Missing one of these disqualifies an otherwise-finished game.
- Prep the boring parts in advance (allowed by most jams): empty project that builds
and exports, input + a title/end screen stub, an export pipeline you've run once, and a
font/SFX source you're licensed to use. Day-one time is too precious to spend here.
- Theme → one sentence. Brainstorm 10 ideas in 15 minutes, then commit to one
expressible as: "You [verb] to [goal] while [constraint]." If you can't say it in one
sentence, it's too big.
- Scope to the clock, not the idea. Use the budget table below. Pick one core
mechanic and one "hook". Everything else is a stretch goal.
- Build the vertical slice first. Get a 30-second playable loop (start → play → lose/win
→ restart) running early. A complete tiny loop beats a half-built big one.
- Reserve the last ~20% of the clock for shipping, not features. Export, test the build
on a clean path, capture screenshots, write the page. Builds always break at hour 47.
- Submit early, update if time remains. Upload a working build well before the
deadline; you can usually replace the file until submissions close. After that, many jams
lock uploads for the whole rating period or allow only declared bug fixes, so the build
you have at the deadline is the one people rate. A submitted mediocre game scores; an
unsubmitted great one does not.
Patterns
1. Scope budget by jam length (plan backwards from the deadline)
| 48-hour jam | 72-96 hour jam | 1-week jam
--------------|------------------------|-----------------------|------------------------
Core mechanic | 1, proven by hour 6 | 1-2 | 2-3 interacting
Levels/content| 1 hand-made or gen'd | 3-5 | 8-12 + progression
Art | placeholder/1 palette | cohesive 1-artist set | themed set + UI
Audio | 3-5 SFX + 1 loop | SFX + 1-2 tracks | adaptive music
SHIP BUFFER | last 8-10 h | last 12-16 h | last full day
2. The 48-hour timeline (concrete blocks)
H0-2 Ideate -> lock ONE-sentence concept -> name the core mechanic + win/lose.
H2-8 Core loop in code with primitives (boxes/circles). Make it playable, no art.
H8-12 Sleep. (Tired code is tomorrow's bug list.)
H12-24 Content: 1 level/encounter, tune difficulty, add the "hook" feature.
H24-30 Sleep + playtest with one other person; cut anything not landing.
H30-38 Art + audio pass. Juice: screenshake, hit-stop, tween, particles, SFX.
H38-44 Bug-fix freeze: no new features. Fix only crashes + soft-locks.
H44-48 EXPORT, test the build clean, screenshots, write page, SUBMIT (by ~H46).
3. Feature triage — decide fast, decide out loud
For every feature ask, in order:
1. Does the core loop work WITHOUT it? -> if yes, it's a stretch goal, not MVP.
2. Can a player tell it's missing? -> if no, cut it.
3. Is it < 30 min of work? -> if no, defer past the ship buffer.
Default answer under deadline pressure is CUT. You can always add in a post-jam version.
4. Pre-submission build check (run on a clean copy, not your dev folder)
[ ] Build runs from an extracted zip on a path with NO engine/IDE installed.
[ ] Controls are shown on-screen or on the page (jurors won't read your mind).
[ ] No dev console errors; no soft-lock; restart works.
[ ] Web build (if any) is set to the platform's playable-in-browser mode.
[ ] Page has: 1-line pitch, controls, screenshots, credits + asset licenses.
[ ] File uploaded and the submission is actually attached to the JAM, not just the page.
5. Format rules of the big recurring jams (re-check the current event page)
Ludum Dare Two categories. COMPO: solo, 48 h, all art/music/sound made during the
event, source code must be shared. JAM: 72 h, solo or team, any base code;
third-party or old assets are fine but opt out of Graphics/Audio voting.
One extra "submission hour" for packaging; after it, only declared bug
fixes and ports, no new features. Entries live on ldjam.com and link to
builds you host elsewhere (for example on itch.io). Rating others raises
your entry's visibility ("Coolness").
GMTK Jam Hosted on itch.io, solo or team, runs about four days. Must be playable
on Windows or in a browser with keyboard/mouse. Generative AI art and
audio get the entry disqualified. Uploads lock during the rating period;
linking to a "fixed" build elsewhere is also grounds for disqualification.
Global Game Run at local sites; each site picks its own 48 h window inside a 7-day
Jam period, and the submission deadline is local time. Submit on
globalgamejam.org with source code and all original assets before the jam
ends; GGJ projects are released open source under a Creative Commons
license, so plan any commercial follow-up with that in mind.
Pitfalls
- Scope creep is the #1 jam killer. If the core loop isn't playable by the first third
of the clock, cut features now, not later.
- No ship buffer. Exporting, zipping, and writing the page reliably takes 1-3 hours and
always surfaces a build bug. Treat the deadline as ~2 hours earlier than it is.
- Submitting to the page but not the jam. On itch.io a jam entry is a separate
submission linked from the jam page — uploading the game alone does not enter it.
- Fixing the build after the deadline. Where uploads lock for rating, the locked build is
what gets rated; posting a link to a newer build can disqualify the entry. Where fixes are
allowed (Ludum Dare), fix only bugs, add no features, and list the changes on the entry.
- Untested export. "Works in the editor" is not "works in the build." Test the exported
artifact on a clean path; missing assets and wrong working directories are classic.
- Unlicensed assets. Music/fonts/sprites pulled from the web can violate jam rules and
break later distribution. Use assets you're licensed for and list them in the credits.
Check the jam's generative AI rule too: some jams disqualify AI-made art or audio.
- Polishing before the loop is fun. Juice multiplies fun; it can't create it. Get the
loop right with placeholders first.
References
- For the throwaway-vs-keep prototyping mindset and greyboxing technique, read the
prototype-fast skill.
- For the actual upload mechanics (project page, channels,
butler push), read
itch-publish.
- Primary sources for Pattern 5: the Ludum Dare rules (
ldjam.com/events/ludum-dare/rules),
the current GMTK Game Jam page on itch.io, and globalgamejam.org ("What is a game jam?"
and "Upload your game!"). Rules change between events; read the live page.
Related skills
prototype-fast — validate a mechanic quickly before committing jam hours to it.
itch-publish — create the page and upload the build (most jams are hosted on itch.io).
- Engine cores (
phaser-core, love2d-core, godot-gdscript, …) and genre skills
(platformer, roguelike, …) — build the actual loop the jam scopes around.
1---2name: game-jam3description: >4 Plan and ship a game under a jam deadline: lock scope to the clock, schedule the hours,5 cut features, and submit on time. Use for a game jam (Ludum Dare, GMTK Jam, Global Game6 Jam), a 48-hour or weekend build, or scoping and submitting a jam entry.7metadata:8 source: https://skillmd.com/skills/junmystery/game-jam9---1011# Game Jam1213Turn a theme and a fixed deadline into a finished, submitted game. This is a planning and14scope-control playbook, not engine code: the win condition is *something submitted and15playable*, not *the game you imagined*.1617## When to use1819- Use when entering a timed jam (Ludum Dare, GMTK Jam, Global Game Jam, a weekend jam),20 scoping a 48-hour build, or deciding what to cut to hit a deadline.21- Use when the user asks "what can I actually build in a weekend" or "how do I submit to22 this jam".2324**When *not* to use:** building the *core loop* in engine code (use the engine skill —25`godot-2d-movement`, `phaser-core`, etc. — or a genre skill like `platformer`); throwaway26experiments with no deadline (use `prototype-fast`); shipping a commercial release (use27`steam-publish` / `itch-publish`).2829## Core workflow30311. **Read the rules before the theme drops.** Confirm the jam's *length*, *theme reveal32 time*, *submission deadline (with timezone)*, whether teams/pre-made assets/engines are33 allowed, its *generative AI* policy, which platforms the build must run on (often Windows34 or browser), *where* you submit (an itch.io jam page, or the jam's own site), whether35 uploads lock once the deadline passes, and whether ratings require you to rate other36 entries. Missing one of these disqualifies an otherwise-finished game.372. **Prep the boring parts in advance** (allowed by most jams): empty project that builds38 and exports, input + a title/end screen stub, an export pipeline you've run once, and a39 font/SFX source you're licensed to use. Day-one time is too precious to spend here.403. **Theme → one sentence.** Brainstorm 10 ideas in 15 minutes, then commit to **one**41 expressible as: *"You [verb] to [goal] while [constraint]."* If you can't say it in one42 sentence, it's too big.434. **Scope to the clock, not the idea.** Use the budget table below. Pick **one** core44 mechanic and **one** "hook". Everything else is a stretch goal.455. **Build the vertical slice first.** Get a 30-second playable loop (start → play → lose/win46 → restart) running early. A complete tiny loop beats a half-built big one.476. **Reserve the last ~20% of the clock for shipping**, not features. Export, test the build48 on a clean path, capture screenshots, write the page. Builds always break at hour 47.497. **Submit early, update if time remains.** Upload a working build *well before* the50 deadline; you can usually replace the file until submissions close. After that, many jams51 lock uploads for the whole rating period or allow only declared bug fixes, so the build52 you have at the deadline is the one people rate. A submitted mediocre game scores; an53 unsubmitted great one does not.5455## Patterns5657### 1. Scope budget by jam length (plan backwards from the deadline)5859```text60 | 48-hour jam | 72-96 hour jam | 1-week jam61--------------|------------------------|-----------------------|------------------------62Core mechanic | 1, proven by hour 6 | 1-2 | 2-3 interacting63Levels/content| 1 hand-made or gen'd | 3-5 | 8-12 + progression64Art | placeholder/1 palette | cohesive 1-artist set | themed set + UI65Audio | 3-5 SFX + 1 loop | SFX + 1-2 tracks | adaptive music66SHIP BUFFER | last 8-10 h | last 12-16 h | last full day67```6869### 2. The 48-hour timeline (concrete blocks)7071```text72H0-2 Ideate -> lock ONE-sentence concept -> name the core mechanic + win/lose.73H2-8 Core loop in code with primitives (boxes/circles). Make it playable, no art.74H8-12 Sleep. (Tired code is tomorrow's bug list.)75H12-24 Content: 1 level/encounter, tune difficulty, add the "hook" feature.76H24-30 Sleep + playtest with one other person; cut anything not landing.77H30-38 Art + audio pass. Juice: screenshake, hit-stop, tween, particles, SFX.78H38-44 Bug-fix freeze: no new features. Fix only crashes + soft-locks.79H44-48 EXPORT, test the build clean, screenshots, write page, SUBMIT (by ~H46).80```8182### 3. Feature triage — decide fast, decide out loud8384```text85For every feature ask, in order:86 1. Does the core loop work WITHOUT it? -> if yes, it's a stretch goal, not MVP.87 2. Can a player tell it's missing? -> if no, cut it.88 3. Is it < 30 min of work? -> if no, defer past the ship buffer.89Default answer under deadline pressure is CUT. You can always add in a post-jam version.90```9192### 4. Pre-submission build check (run on a clean copy, not your dev folder)9394```text95[ ] Build runs from an extracted zip on a path with NO engine/IDE installed.96[ ] Controls are shown on-screen or on the page (jurors won't read your mind).97[ ] No dev console errors; no soft-lock; restart works.98[ ] Web build (if any) is set to the platform's playable-in-browser mode.99[ ] Page has: 1-line pitch, controls, screenshots, credits + asset licenses.100[ ] File uploaded and the submission is actually attached to the JAM, not just the page.101```102103### 5. Format rules of the big recurring jams (re-check the current event page)104105```text106Ludum Dare Two categories. COMPO: solo, 48 h, all art/music/sound made during the107 event, source code must be shared. JAM: 72 h, solo or team, any base code;108 third-party or old assets are fine but opt out of Graphics/Audio voting.109 One extra "submission hour" for packaging; after it, only declared bug110 fixes and ports, no new features. Entries live on ldjam.com and link to111 builds you host elsewhere (for example on itch.io). Rating others raises112 your entry's visibility ("Coolness").113GMTK Jam Hosted on itch.io, solo or team, runs about four days. Must be playable114 on Windows or in a browser with keyboard/mouse. Generative AI art and115 audio get the entry disqualified. Uploads lock during the rating period;116 linking to a "fixed" build elsewhere is also grounds for disqualification.117Global Game Run at local sites; each site picks its own 48 h window inside a 7-day118Jam period, and the submission deadline is local time. Submit on119 globalgamejam.org with source code and all original assets before the jam120 ends; GGJ projects are released open source under a Creative Commons121 license, so plan any commercial follow-up with that in mind.122```123124## Pitfalls125126- **Scope creep is the #1 jam killer.** If the core loop isn't playable by the first third127 of the clock, cut features now, not later.128- **No ship buffer.** Exporting, zipping, and writing the page reliably takes 1-3 hours and129 always surfaces a build bug. Treat the deadline as ~2 hours earlier than it is.130- **Submitting to the page but not the jam.** On itch.io a jam entry is a separate131 *submission* linked from the jam page — uploading the game alone does not enter it.132- **Fixing the build after the deadline.** Where uploads lock for rating, the locked build is133 what gets rated; posting a link to a newer build can disqualify the entry. Where fixes are134 allowed (Ludum Dare), fix only bugs, add no features, and list the changes on the entry.135- **Untested export.** "Works in the editor" is not "works in the build." Test the exported136 artifact on a clean path; missing assets and wrong working directories are classic.137- **Unlicensed assets.** Music/fonts/sprites pulled from the web can violate jam rules and138 break later distribution. Use assets you're licensed for and list them in the credits.139 Check the jam's generative AI rule too: some jams disqualify AI-made art or audio.140- **Polishing before the loop is fun.** Juice multiplies fun; it can't create it. Get the141 loop right with placeholders first.142143## References144145- For the throwaway-vs-keep prototyping mindset and greyboxing technique, read the146 `prototype-fast` skill.147- For the actual upload mechanics (project page, channels, `butler push`), read148 `itch-publish`.149- Primary sources for Pattern 5: the Ludum Dare rules (`ldjam.com/events/ludum-dare/rules`),150 the current GMTK Game Jam page on itch.io, and `globalgamejam.org` ("What is a game jam?"151 and "Upload your game!"). Rules change between events; read the live page.152153## Related skills154155- `prototype-fast` — validate a mechanic quickly before committing jam hours to it.156- `itch-publish` — create the page and upload the build (most jams are hosted on itch.io).157- Engine cores (`phaser-core`, `love2d-core`, `godot-gdscript`, …) and genre skills158 (`platformer`, `roguelike`, …) — build the actual loop the jam scopes around.