Prototype

/prototype — Rapid Prototyping Workflow

Delegate to: luau-gameplay-programmer or luau-systems-programmer (depending on scope)

Purpose

Prototypes are throwaway code to test an idea quickly. They are NOT production code.

Rules

• Prototypes live in prototypes//
• Prototypes NEVER import from src/
• Prototype code does NOT need to meet production standards
• Prototypes require a README.md explaining the hypothesis
• Prototypes are NOT graduated directly to src/ — if the idea is good, rewrite cleanly

Steps

1. Define the Hypothesis

Ask the user:

• "What are we testing?"
• "What's the question this prototype answers?"
• "What would 'success' look like?"
• "What's the time budget?" (default: 2-4 hours)

2. Scope Ruthlessly

• Only build what's needed to test the hypothesis
• Skip: polished UI, data persistence, multiplayer (unless core to hypothesis)
• Focus: the thing that's being questioned

3. Set Up Prototype Directory

prototypes/
└── /
    ├── README.md        # Hypothesis, how to run, findings
    ├── src/             # Prototype Luau files
    └── assets/          # Any placeholder assets

4. Build the Minimum

Write code that answers the question. Do NOT:

• Add type annotations (unless they help)
• Wrap in pcall (unless necessary for the question)
• Write tests
• Worry about cleanup
• Optimize performance

DO:

• Use clear variable names (so future-you understands)
• Comment the KEY decisions
• Make it runnable (Rojo or manual Studio import)

5. Document Findings

After running, update the README:

# Prototype: [Name]

## Hypothesis
[What we were testing]

## How to Run
1. [Setup steps]
2. [Run steps]

## What We Learned
- [Finding 1]
- [Finding 2]
- [Unexpected result]

## Go / No-Go
**Verdict**: [Good idea / bad idea / needs more research]

## If Go, Next Steps
1. [How to translate this into production]
2. [What specialists are needed]
3. [What systems it affects]

## If No-Go
- [Why it didn't work]
- [Alternative approaches to consider]

6. Decide

User decides:

• Good idea: Schedule proper implementation (rewrite, not copy-paste)
• Good idea with tweaks: Iterate the prototype OR move to production planning
• Bad idea: Document learnings, archive the prototype, move on
• Inconclusive: Decide if more prototyping is worth the time

Anti-Patterns

• Prototype sprawl (spending weeks on a "prototype")
• Importing prototype code into production without rewrite
• Skipping the README and losing the findings
• Testing multiple hypotheses in one prototype (hard to learn from)

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: CodePhobiia
• Source: CodePhobiia/claude-roblox-game-studio
• License: MIT

Install and usage instructions live in the source repository linked above.