everything-game-dev-code
AGENTS.md105 starsChanged 8 months agoMITView on GitHub
If a $imagegen output becomes a real project asset, move the accepted file into the workspace and keep the naming and folder conventions defined by the relevant art pipeline skill.
Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.
Of the AGENTS.md files we read, the 31,843 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.
In this repository117 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- .github/copilot-instructions.mdCopilot instructions
- .cursor/rules/00-project-context.mdcCursor rule
- .cursor/rules/01-engine-isolation.mdcCursor rule
- .cursor/rules/02-workflow.mdcCursor rule
- .cursor/rules/03-doc-sync.mdcCursor rule
- .cursor/rules/README.mdCursor rule
- 2d-animation-pipelineSkill
- 3d-animation-pipelineSkill
- 3d-asset-pipelineSkill
- ai-asset-generationSkill
- art-bibleSkill
- audio-implementationSkill
- blender-environment-pipelineSkill
- cinematic-pipelineSkill
- dialogue-content-pipelineSkill
- generated-raster-asset-pipelineSkill
- lighting-lod-pipelineSkill
- localization-pipelineSkill
- materials-shading-pipelineSkill
- placeholder-asset-pipelineSkill
- rigging-skinning-pipelineSkill
- sprite-pipelineSkill
- technical-art-pipelineSkill
- tilemap-pipelineSkill
- ui-animation-pipelineSkill
- ui-asset-pipelineSkill
- vfx-pipelineSkill
- accessibility-designSkill
- combat-designSkill
- core-loop-designSkill
- economy-balancingSkill
- game-feel-designSkill
- level-designSkill
- liveops-designSkill
- monetization-designSkill
- narrative-designSkill
- onboarding-tutorial-designSkill
- procgen-designSkill
- progression-designSkill
- quest-designSkill
- ai-behavior-patternsSkill
- animation-state-patternsSkill
- asset-managementSkill
- build-pipeline-patternsSkill
- event-bus-patternsSkill
- gameplay-architectureSkill
- input-abstractionSkill
- memory-budgetingSkill
- multiplayer-netcode-patternsSkill
- performance-budgetingSkill
- physics-gameplay-patternsSkill
- save-system-patternsSkill
- telemetry-instrumentationSkill
- tools-pipeline-patternsSkill
- ui-hud-patternsSkill
- godot-3d-scenesSkill
- godot-build-releaseSkill
- godot-csharp-patternsSkill
- The first 60 shown.
The file
Show its rules and text
We split this AGENTS.md into the rules it gives an agent
16 rules. About 0 are common, 4 are distinctive, and 12 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈4
Distinctive · estimate
This author's own choices. Worth a look.
12
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 9
Use engine-neutral guidance by default.
- Distinctive: line 10
Only load one engine-specific rule/skill family at a time unless the task is explicitly comparative research.
- Very specific: line 13
Start from commands/ if the task maps to a named workflow.
- Very specific: line 14
Use agents/ to pick role ownership.
- Very specific: line 15
Use skills/ for repeatable execution patterns.
- Very specific: line 16
Use rules/common/ plus one engine layer if implementation is engine-specific.
- Distinctive: line 17
Keep docs, QA, telemetry, and release implications explicit.
- Very specific: line 20
When a user types a scaffold command such as /plan, /verify, or /unity-review, resolve it to commands/<name>.md, read that file, and …
- Very specific: line 23
Use .codex/config.toml profiles for scaffold-oriented model and approval defaults.
- Very specific: line 24
Use codex review for code-review workflows when the user asks for a review.
- Very specific: line 25
Use codex exec for repeatable non-interactive validation or automation when appropriate.
- Very specific: line 26
Use codex mcp for external tool servers described by mcp-configs/.
- Very specific: line 27
Use $imagegen for raster-first asset generation or bitmap edits when the task is better served by PNG/WebP-style output than by repo-native …
- Very specific: line 28
Use Codex skills or plugins only as packaging layers over the shared skills/ source of truth.
- Very specific: line 30
If a $imagegen output becomes a real project asset, move the accepted file into the workspace and keep the naming and folder conventions …
- Distinctive: line 33
Never mix Unity, Unreal, Godot, and web implementation advice in the same production task.
≈ “Common” is an estimate: at least 1 in 5 of the 40,246 AGENTS.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
AGENTS.md for Codex
Project purpose
This repository is a game-development agent scaffold with shared layers for rules, agents, commands, skills, contexts, hooks, and harness adapters.
Instruction hierarchy
- Root-level project guidance belongs in this file and in the repo's main
AGENTS.md. - Shared workflow logic lives in
agents/,commands/,skills/, andrules/. - Use engine-neutral guidance by default.
- Only load one engine-specific rule/skill family at a time unless the task is explicitly comparative research.
How to work in this repo
- Start from
commands/if the task maps to a named workflow. - Use
agents/to pick role ownership. - Use
skills/for repeatable execution patterns. - Use
rules/common/plus one engine layer if implementation is engine-specific. - Keep docs, QA, telemetry, and release implications explicit.
Command handling
When a user types a scaffold command such as /plan, /verify, or /unity-review, resolve it to commands/<name>.md, read that file, and execute the workflow it defines. Treat .codex/commands/ files as thin compatibility wrappers only.
Codex surfaces
- Use
.codex/config.tomlprofiles for scaffold-oriented model and approval defaults. - Use
codex reviewfor code-review workflows when the user asks for a review. - Use
codex execfor repeatable non-interactive validation or automation when appropriate. - Use
codex mcpfor external tool servers described bymcp-configs/. - Use
$imagegenfor raster-first asset generation or bitmap edits when the task is better served by PNG/WebP-style output than by repo-native code, SVG, or procedural placeholders. - Use Codex skills or plugins only as packaging layers over the shared
skills/source of truth.
If a $imagegen output becomes a real project asset, move the accepted file into the workspace and keep the naming and folder conventions defined by the relevant art pipeline skill.
Engine isolation
Never mix Unity, Unreal, Godot, and web implementation advice in the same production task.
Expected outputs
- plans with owners and risks
- design and technical docs when needed
- implementation that respects the chosen engine layer
- verification notes and doc updates when behavior changes
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
Reviews can't be read right now.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.