VibeGame
CLAUDE.md91 starsChanged 12 months agoMITView on GitHub
Required: Read layers/structure.md before proceeding with any task
In this repositoryOne other file it gives its agents.
- llms.txtllms.txt
The file
We split this CLAUDE.md into the rules it gives an agent
3 rules. About 0 are common, 3 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 7
Required: Read layers/structure.md before proceeding with any task
- Distinctive: line 20
Simplicity: Keep code simple, elegant, concise, and readable
- Distinctive: line 21
Structure: Keep files small and single-responsibility.
≈ “Common” is an estimate: at least 1 in 5 of the 27,641 CLAUDE.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
AI Context - Working Agreement
<project-description>
VibeGame - A vibe coding game engine using an ECS architecture with bitECS, featuring a Bevy-inspired plugin system and A-frame-style declarative XML recipes.
</project-description>Required: Read layers/structure.md before proceeding with any task
Context Management System
- Tier 0 — global:
CLAUDE.md(root). Global standards and system overview - Tier 1 — project:
layers/structure.md. Project map (stack, commands, layout, entry points) - Tier 2 — folder context:
context.mdin any folder; one per folder; explains purpose/structure of that folder - Tier 3 — implementation: Code files (scripts)
Rules
- Priority: Your number one priority is to manage your own context; always load appropriate context before doing anything else
- No History: CRITICAL - Code and context must NEVER reference their own history. Write everything as the current, final state. Never include comments like "changed from X to Y" or "previously was Z". This is a severe form of context rot
- Simplicity: Keep code simple, elegant, concise, and readable
- Structure: Keep files small and single-responsibility; separate concerns (MVC/ECS as appropriate)
- Reuse: Reuse before adding new code; avoid repetition
- Comments: Code should be self-explanatory without comments; use concise comments only when necessary
- State: Single source of truth; caches/derivations only
- Data: Favor data-driven/declarative design
- Fail Fast: Make bugs immediately visible rather than hiding them; favor simplicity over defensive patterns
- Backwards Compatibility: Unless stated otherwise, favor simplicity over backwards compatibility; the design rules above should make breaking changes easy to trace and fix
Security
- Inputs & secrets: Validate inputs; secrets only in env; never log sensitive data
- Auth: Gateway auth; server-side token validation; sanitize inputs
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
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.

