Roxlit - Connect AI Coding Tools to Roblox Studio
Roxlit is an open-source launcher that connects Claude Code, Cursor, and Windsurf to Roblox Studio via Rojo and an MCP server, with Luau context packs for accurate game code generation.
TL;DR
TL;DR: Roxlit is an open-source launcher that connects AI coding tools like Claude Code, Cursor, and Windsurf to Roblox Studio using Rojo for file sync and an MCP server for runtime Luau execution, plus curated context packs that teach the AI to write correct Roblox code.
Source and Accuracy Notes
All links verified from actual source (github.com/Roxlit/installer, README.md checked 2026-08-12):
- Project page: roxlit.dev
- Source repository: github.com/Roxlit/installer (MIT License, verified)
- HN launch thread: news.ycombinator.com/item?id=47184387
What Is Roxlit?
Roxlit is an open-source launcher purpose-built for Roblox game developers who want to use AI coding assistants inside their workflow. Without it, AI tools like Claude Code and Cursor generate poor Luau code — they use deprecated APIs like wait() instead of task.wait(), put logic in the wrong script types, and hallucinate Roblox services that do not exist.
The project connects AI tools to Roblox Studio in two ways:
File sync via Rojo — Roxlit launches rojo serve to sync files from the AI editor into Roblox Studio in real time. Scripts, instances, models, and properties all update live in the game hierarchy.
Runtime access via MCP server — The AI executes Luau code directly inside Studio through an MCP server. It can create Parts, build GUIs, run playtests, query the game hierarchy, and read properties. The AI is no longer blind — it sees and manipulates the game while you build.
Roxlit also bundles 7 AI context packs that teach your AI how to write correct Luau code. These are curated documentation files installed into .roxlit/context/, covering Roblox services, Luau data types, client-server architecture, DataStores, RemoteEvents, physics, and networking. The AI reads these automatically and produces code that follows Roblox best practices.
Setup Workflow
Prerequisites
- Windows 10/11 (x64) — currently Windows only
- Roblox Studio installed
- Node.js (for MCP server)
- A Roblox place open in Studio
Step 1: Install Roxlit
Download the latest release from roxlit.dev or the GitHub Releases page. Run the installer wizard.
The first-run wizard asks you to:
- Pick your AI tool (Claude Code, Cursor, Windsurf, or any editor that supports MCP)
- Select your Roblox project folder
- Roxlit detects what is already installed and what is missing
- One click installs everything (Rojo, MCP server, context packs)
Step 2: Connect Your AI Editor
After installation, the Roxlit launcher window stays open. Click Start to launch:
rojo serve(file sync server)- Your chosen editor (Claude Code, Cursor, etc.)
- The MCP server for Studio
Studio must have the Roxlit plugin installed and active. The plugin auto-connects when you start development.
Step 3: Code with AI
With the launcher running, open your editor. Changes you make to .lua files in your project folder sync into Studio via Rojo in real time. For runtime actions (creating instances, running playtests, reading game state), the AI calls the MCP server, which executes Luau inside Studio and streams console output back to your session.
How the Context Packs Work
The context packs are the core differentiator. Without them, AI tools write incorrect Roblox code because their training data does not emphasise Roblox-specific patterns. Roxlit’s packs fix this by installing curated reference files into your project:
.roxlit/
context/
services.md # Roblox services and correct API usage
data-types.md # Luau strict typing patterns
client-server.md # Security model and architecture
datastores.md # DataStore patterns and error handling
remote-events.md # Client-server communication
physics.md # Constraints and spatial queries
networking.md # Replication and networking
The AI reads these files when working in your project and produces Luau that follows Roblox conventions.
Platform Support
| Editor | Status | |---|---| | Claude Code | Supported | | Cursor | Supported | | Windsurf | Supported | | Any MCP-compatible editor | Supported via MCP |
Windows x64 only. No macOS or Linux build at this time.
Security Notes
- The MCP server executes Luau inside your running Roblox Studio session. Use it only in your own studio sessions, not on shared or production game servers.
- Context packs are local files in your project directory. They do not send data to external servers.
- The installer is not code-signed. Windows SmartScreen shows a warning on first run.
FAQ
Q: Does this work with Roblox Studio on macOS? A: No. Roxlit currently supports Windows x64 only.
Q: Can I use this with VS Code? A: Any editor that supports MCP can connect. VS Code has an MCP extension. The README does not list it explicitly, but the MCP protocol is editor-agnostic.
Q: Does Roxlit work with existing Rojo projects?
A: Yes. Roxlit detects an existing default.project.json and connects to your current project setup rather than replacing it.
Q: What happens if the AI writes bad Luau code? A: The context packs significantly reduce common errors. However, always review AI-generated code before running playtests in shared experiences — the AI can still produce logic errors.
Conclusion
Roxlit fills a specific gap: Roblox developers who want AI assistance without the frustration of AI-generated Luau that uses deprecated APIs or wrong script types. The combination of Rojo file sync, an MCP server for live Studio access, and purpose-built context packs makes the workflow coherent. It is a one-person project with a small community, but the core loop works and the source is MIT-licensed on GitHub.
If you develop Roblox games on Windows and want to integrate Claude Code, Cursor, or similar tools into your workflow, Roxlit is worth trying. Check the GitHub releases for the latest build.
Related Posts
ai-setup
Recall – Persistent Memory for Claude Code via MCP Hooks
Recall gives Claude Code a permanent memory store that survives session restarts and context compaction. Four hooks capture and restore context automatically — with cloud SaaS or self-hosted options.
2/28/2026
dev-tools
AgentMesh – Define AI Agent Teams in YAML
Define multi-agent AI workflows in YAML and run them locally with one command. AgentMesh brings Docker Compose patterns to AI agent orchestration.
5/28/2026
ai-setup
Sentrial – Catch AI Agent Failures Before Your Users Do
YC W26-backed AI agent observability platform. Trace sessions, detect silent regressions, and A/B test prompts in production before failures reach users.
5/28/2026