agentleFS
EL4CTEOEL4CTEO/roblox-skills/skills/roblox-code-review/SKILL.md

roblox-code-review

Skill92 starsChanged 47 days agoMITView on GitHub

Review Roblox/Luau code for exploits, data loss, memory leaks, performance problems, lifecycle bugs, policy issues, and deprecated or superseded APIs, using a severity-ranked checklist and the full 2026 deprecation map. Use when reviewing a PR or script, auditing a game before launch, modernizing legacy code, or asked what is wrong with some Roblox code.

2 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

6 rules. About 0 are common, 2 are distinctive, and 4 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 4

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Verify claims against the code (and the engine reference when unsure) — don't pattern-match blindly.

    line 9
  2. Very specific:

    Every OnServerEvent/OnServerInvoke: types checked with typeof, numbers math.isfinite, strings length/UTF-8 checked, instances IsA + …

    line 23
  3. Very specific:

    No server-only logic, admin lists, or secrets in ReplicatedStorage/StarterPlayer/Workspace.

    line 25
  4. Distinctive:

    No yields at module top level.

    line 43
  5. Very specific:

    No per-frame FindFirstChild/GetDescendants/Instance.new.

    line 48
  6. Very specific:

    All player-authored text shown to others filtered server-side (FilterStringAsync), fail closed.

    line 54

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill 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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Roblox code review

Review in severity order. Report each finding with file:line, severity, what breaks, and the fix. Verify claims against the code (and the engine reference when unsure) — don't pattern-match blindly.

Severity guide

SeverityExamples
CriticalClient-authoritative rewards/damage/currency; unvalidated remote args (NaN, instance spoofing); purchases granted outside ProcessReceipt or non-idempotent; data saved with SetAsync from multiple servers without session locking; secrets/API keys in replicated code; unfiltered player text shown to others.
HighData loss paths (no BindToClose, saving default data after a failed load, no retry); memory leaks (connections, player-keyed tables); InvokeClient; per-frame remotes; infinite yields (WaitForChild without timeout on optional instances, RemoteFunction handlers that can yield forever); streaming-unsafe client indexing.
MediumDeprecated/superseded APIs; wait/spawn/delay; polling loops; server-side tweens; missing pcall on network/yielding APIs; missing late-joiner handling; race conditions on CharacterAdded.
LowStyle, naming, missing types, magic numbers, dead code, minor inefficiencies.

Checklist

Security (roblox-security)

  • Every OnServerEvent/OnServerInvoke: types checked with typeof, numbers math.isfinite, strings length/UTF-8 checked, instances IsA + IsDescendantOf(trusted), rate limited, context checked.
  • No server-only logic, admin lists, or secrets in ReplicatedStorage/StarterPlayer/Workspace.
  • ProximityPrompt/ClickDetector/Touched handlers re-validate distance/state on the server.
  • Gameplay-critical unanchored parts are server-owned.

Data (roblox-data-stores)

  • Session locking (ProfileStore or equivalent); UpdateAsync for read-modify-write; retries with backoff.
  • BindToClose flushes saves in parallel; no saves of Instance/Vector3/mixed tables.
  • Failed loads never overwritten with defaults; player-keyed tables cleared on leave.
  • Keys stable (User_{UserId}), Studio uses a separate store/universe.

Monetization (roblox-monetization)

  • Single ProcessReceipt; idempotent by PurchaseId; returns PurchaseGranted only after persisting.
  • Pass perks checked on join and after in-session purchase; prices not hardcoded; paid random items disclose odds and respect PolicyService.

Correctness & lifecycle (roblox-architecture)

  • Handles players already in game (for _, p in Players:GetPlayers()) and characters already spawned.
  • Per-character state rebuilt on respawn; per-player state cleaned on PlayerRemoving.
  • No yields at module top level; no circular requires; one owner per piece of state.
  • Deferred-event safe (doesn't rely on handlers running synchronously).
  • Client code streaming-safe (WaitForChild/nil checks for Workspace content).

Performance (roblox-performance)

  • No per-frame FindFirstChild/GetDescendants/Instance.new; params objects reused.
  • Event-driven instead of polling; think loops throttled and staggered.
  • Tweens and cosmetic effects on clients; remotes batched/throttled; payloads minimal.
  • Connections disconnected / instances destroyed; no unbounded caches.

Text & safety (roblox-text-chat)

  • All player-authored text shown to others filtered server-side (FilterStringAsync), fail closed.
  • Chat uses TextChatService; public editable text rate-limited (≥ 1 minute).

Modern API usage

  • task.*, *Async variants, Raycast, mover constraints, PivotTo, Animator:LoadAnimation, TextChatService, UIDragDetector, workspace collision-group APIs, CreatePath+ComputeAsync.
  • Full map of deprecated and superseded members: references/deprecated-apis.md.

Code quality (roblox-luau)

  • --!strict with typed public APIs; services via GetService; no _G/shared.
  • Clear module boundaries; config separated from logic; tests for pure logic (roblox-testing).

Output format

## Summary
<2–3 sentences: overall risk, the most important fixes>

## Findings
1. [Critical] src/server/Shop.server.luau:42 — Price read from client argument.
   Impact: exploiters buy anything for 0 coins. Fix: look up ITEMS[itemId].price on the server.
2. [High] ...

## Modernization (optional)
- Replace `wait()` → `task.wait()` (12 occurrences) ...

Offer to apply fixes; when fixing, keep behavior identical except for the bug, and re-run the relevant checks (type check, tests, Studio playtest via roblox-studio-mcp).

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.