roblox-debug
Skill1 starsChanged 4 months agoMITView on GitHub
Use when diagnosing Roblox or Luau errors, crashes, or unexpected behavior with an iterative reproduce, inspect, fix, and verify loop.
In this repository32 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- roblox-analyticsSkill
- roblox-animated-modelsSkill
- roblox-animation-vfxSkill
- roblox-architectureSkill
- roblox-assetsSkill
- roblox-audioSkill
- roblox-buildingSkill
- roblox-cameraSkill
- roblox-cloudSkill
- roblox-code-reviewSkill
- roblox-dataSkill
- roblox-guiSkill
- roblox-inputSkill
- roblox-lightingSkill
- roblox-localizationSkill
- roblox-luau-coreSkill
- roblox-luau-patternsSkill
- roblox-luau-typesSkill
- roblox-monetizationSkill
- roblox-networkingSkill
- roblox-npc-aiSkill
- roblox-oauthSkill
- roblox-performanceSkill
- roblox-physicsSkill
- roblox-publish-checklistSkill
- roblox-securitySkill
- roblox-server-dataSkill
- roblox-sharp-edgesSkill
- roblox-studio-mcpSkill
- roblox-toolingSkill
- roblox-ui-designSkill
The file
Files
When to Load
Load when diagnosing errors, crashes, or unexpected behavior in Roblox/Luau code. Provides a structured 7-step iterative debug loop with a max of 5 iterations before escalating. Works in both file-sync and MCP modes.
Quick Reference
7-Step Debug Loop (max 5 iterations)
1. Error Gathering — Get exact error message, stack trace (script + line), trigger condition (join/action/timer). Use MCP for console output if available.
2. Code Discovery — Read relevant script(s). Search for error text, function names, or script from stack trace.
3. Root Cause Analysis — Categorize: Syntax (missing end) · Runtime (nil, destroyed Instance) · Logic (wrong calc/condition) · Security (client-side logic) · Performance (timeout, memory). Load roblox-sharp-edges for known gotchas. Find the exact line + true root cause.
4. Generate Fix — Produce corrected Luau. Explain what was wrong, why fix works, related code to check. Flag architectural changes.
5. Apply & Test — Write fix to file. MCP verify if available. In offline mode: before/after diff + manual test instructions.
MCP evidence: discover/select Studio, inspect before editing, read back the fix, run playtest when relevant, and report console/screenshot limitations explicitly.
6. Verify — Resolved → Step 7. New errors → back to Step 1. Same error → new hypothesis, back to Step 3. Iteration 5 → escalate with full diagnosis.
7. Summary — Document: bug description · root cause · fix applied · verification · related risks · prevention pattern. If systemic, recommend /code-review.
Escalation (at iteration 5)
Output: all attempted fixes + why rejected, remaining hypotheses, recommended next steps (manual investigation, DevForum search, etc.).
📖 Full reference: references/full.md
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.