
roblox-mcp
Connect an AI client to Roblox Studio through MCP. Use for connection setup, Studio tools, multiple Studio windows, playtesting, or combining MCP with Script Sync.
- Installs
- 321
- Last 24h
- —
- Trending rank
- —
- Security
- Not audited
Install
Works with every agent the skills CLI supports.
npx skills add https://github.com/nonlooped/roblox-suite --skill roblox-mcp- 01
Install
Run the command in your project. The CLI asks which agents to install for.
- 02
Describe the task
Ask your agent for something this skill covers. It loads roblox-mcp automatically when relevant.
- 03
Review the result
The agent follows the SKILL.md steps, scripts and references bundled with the skill.
SKILL.md
The instructions your agent reads when the skill is triggered.
roblox-mcp
Official sources:
- https://create.roblox.com/docs/en-us/studio/mcp
- https://create.roblox.com/docs/en-us/ai/build
- https://create.roblox.com/docs/en-us/scripting/sync
- https://modelcontextprotocol.io/docs/getting-started/intro
The official Roblox Studio MCP server lets a connected AI client inspect and edit an open place. It provides tools for scripts, instances, Luau execution, assets, and playtesting.
What the Studio MCP server is
Roblox Studio implements a Model Context Protocol (MCP) server using stdio transport. The server runs as a local process on your machine and proxies requests from a supported AI client into the active Studio session. It is the bridge between your editor's agent and Studio: Script Sync handles disk-to-Studio file syncing, while MCP handles everything else (Explorer selection, instance inspection, script editing, Luau execution, play mode, input simulation, and asset insertion).
Prerequisites
- Latest Roblox Studio. Update through the launcher or download from the Creator Hub.
- An MCP-capable AI client. Cursor, VS Code with Claude Code, Claude Desktop, Codex CLI, Gemini CLI, Antigravity, or any client that supports
stdiotransport. - A Roblox account signed into Studio.
- Git (recommended) for the file-based workflow.
Enabling the server in Studio
- Open Studio and load the place you want the agent to work on.
- Open Assistant (button in the upper right).
- Click ⋯ → Manage MCP Servers.
- Turn on Enable Studio as MCP server.
- The panel shows quick-connect options and manual configuration snippets. A green indicator appears when a client connects.
If you do not see MCP options, restart Studio after updating to the latest version.
Connecting your client
Choose the method that matches your client:
Quick connect supports Antigravity, Codex CLI, Claude Code, Claude Desktop, Cursor, Gemini CLI, and Visual Studio Code.
- In Studio: Assistant → ⋯ → Manage MCP Servers → Quick connect.
- Turn on your installed client.
- Restart the client if the tools do not appear immediately.
For clients that read an mcp.json or similar config file, see references/setup-and-connection.md for per-OS, copy-paste configurations.
For clients that need a raw CLI command, see references/setup-and-connection.md for Windows and macOS commands.
After connecting, verify with:
Use the Roblox MCP to read the current game tree in Roblox Studio. List what's in Workspace.
What you can do with the MCP tools
The server exposes tools in several categories. The current tool names and usage guidance are in references/tool-reference.md.
| Category | Examples |
|---|---|
| Scripts | script_read, multi_edit, script_search, script_grep |
| Data model | search_game_tree, inspect_instance, subagent (explore) |
| Luau execution | execute_luau |
| Asset generation & insertion | generate_mesh, generate_material, generate_procedural_model, wait_job_finished, search_asset, insert_asset, upload_image, store_image |
| Playtesting | start_stop_play, get_studio_state, get_console_output, screen_capture, subagent (playtest) |
| Input simulation | character_navigation, user_keyboard_input, user_mouse_input |
| Documentation | http_get, skill |
| Instance discovery | list_roblox_studios; pass studio_id on subsequent calls |
Work with MCP and Script Sync
Script Sync maps folders on disk (ServerScriptService/, ReplicatedStorage/, StarterPlayerScripts/, etc.) to Studio services, so .luau files you edit locally appear in Studio automatically. MCP covers everything Script Sync cannot reach: editing StarterGui, inserting models, running commands, playtesting, and inspecting instances.
Typical agent loop:
- Edit disk files through Script Sync for reusable modules and scripts.
- Use MCP to inspect instances, run Luau snippets, start play mode, and capture output/screenshots.
- Use MCP to edit scripts that live in containers not covered by Script Sync (for example,
StarterGui).
See references/script-sync-integration.md for the full setup and a starter-project layout.
Multi-instance handling
Call list_roblox_studios to discover connected windows, their Studio instance IDs, and place IDs. Local places without a place ID are listed by name. Pass the selected instance's studio_id on every subsequent tool call; do not rely on a shared active window or infer the target from an object path. Re-list instances when a window closes or the target changes.
Security and trust
MCP clients can read and modify your open places. Treat MCP connections like any privileged integration:
- Only connect clients you trust.
- Work on test places or version-controlled projects.
- Review agent edits before publishing.
- Keep
.rbxlxand other place files out of Git (use the starter.gitignore). - Disconnect or disable the MCP server when it is not in use.
- Treat
execute_luauas privileged code: it can publish places, write to DataStores, make HTTP requests, and access credentials loaded in Studio.
See references/security-and-troubleshooting.md for the full security checklist and troubleshooting steps.
Verification checklist
- Studio is updated and MCP is enabled.
- AI client shows the Roblox MCP tools after restart.
-
script_readorsearch_game_treereturns the current place structure. - Script Sync maps the expected service folders to disk (if using file-based workflow).
-
execute_luaucan run a simpleprintand return output. -
start_stop_playenters/exits play mode successfully.
Scripts
scripts/MCPReadyChecker.lua: a diagnostic snippet you can run withexecute_luauto verify Script Sync status and basic model health.scripts/StudioModelProbe.lua: a reusable utility for summarizing the game tree, useful as a pattern for agent exploration prompts.
How to proceed
- Confirm prerequisites and enable MCP in Studio.
- Connect your client using quick connect or the manual config in references/setup-and-connection.md.
- Verify the connection with a simple tree-read or
execute_luaucall. - If using a file-based workflow, set up Script Sync per references/script-sync-integration.md.
- Use the tool reference to craft precise agent prompts and the security guide to keep the workflow safe.
Reference index
- script-sync-integration.md: Combine files on disk with Studio MCP edits and playtests.
- security-and-troubleshooting.md: Investigate MCP connection failures or review Studio access.
- setup-and-connection.md: Enable Studio MCP and connect a client on Windows or macOS.
- tool-reference.md: Choose a Studio MCP tool and identify the target Studio instance.
FAQ
What does the roblox-mcp skill do?
Connect an AI client to Roblox Studio through MCP. Use for connection setup, Studio tools, multiple Studio windows, playtesting, or combining MCP with Script Sync.
How do I install roblox-mcp?
Run npx skills add https://github.com/nonlooped/roblox-suite --skill roblox-mcp in your project. The skills CLI detects the agents you use and copies the skill into the right folder. For Claude Code you can also place the skill folder at ~/.claude/skills/roblox-mcp/ (personal) or .claude/skills/roblox-mcp/ (project).
Which AI agents can use roblox-mcp?
Any agent that supports the open Agent Skills format (a folder with a SKILL.md file), including Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, Windsurf, Cline, OpenCode and more.
Is roblox-mcp safe to install?
No partner security audit is available yet. Skills can run commands and scripts, so read the SKILL.md and any bundled files before installing.
Who maintains roblox-mcp?
It is published by nonlooped in nonlooped/roblox-suite. It has 321 recorded installs.
More from nonlooped/roblox-suite
roblox-vfx
nonlooped/roblox-suite
Build or optimize Roblox particles, beams, trails, and highlights. Use for emitter shapes, flipbooks, one-shot bursts, animation-marker effects, distance culling, and overdraw.
roblox-animation
nonlooped/roblox-suite
Build or debug Roblox character animations, IK poses, animation-marker events, and UI or object tweens. Use for locomotion, emotes, motion blending, and timed effects.
roblox-user-interfaces
nonlooped/roblox-suite
Open the skill to read its SKILL.md instructions.
roblox-testing
nonlooped/roblox-suite
Open the skill to read its SKILL.md instructions.
roblox-networking
nonlooped/roblox-suite
Design or review Roblox client-server communication. Use for remotes, bindables, replication, network ownership, input validation, rate limiting, and exploit defenses.
Popular ai & agents skills
google-agents-cli-publish
google/agents-cli
This skill should be used when the user wants to "publish an agent", "publish my ADK agent", "register an agent with Gemini", "publish to Gemini" (formerly Gemini Enterprise), or needs guidance on the agents-cli publish gemini command. Also use when the user wants to "manage agents in Agent Registry", "list/update/delete registered agents", or "register an MCP server". Covers ADK vs A2A registration modes, programmatic and interactive usage, flag reference, auto-detection from deployment metadata, Agent Registry fleet management, and troubleshooting. Part of the agents-cli skills suite. Do NOT use for deployment (use google-agents-cli-deploy).
google-agents-cli-eval
google/agents-cli
This skill should be used when the user wants to "run an evaluation", "evaluate my agent", "evaluate my ADK agent", "write an eval dataset", "analyze eval failures", "compare eval results", "optimize agent", or needs guidance on the Gemini platform eval methodology and the Quality Flywheel. Covers eval metrics, dataset schema, LLM-as-judge scoring, and common failure causes. Applies to any agents-cli project, whatever framework the agent is written in. Do NOT use for agent API code patterns (ADK: use google-agents-cli-adk-code), deployment (use google-agents-cli-deploy), or project scaffolding (use google-agents-cli-scaffold).
skill-creator
anthropics/skills
Create new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy.
codex-pet
prime-skills/runcomfy-agent-skills
Codex Pet generator on RunComfy. Build a Codex-compatible Codex Pet spritesheet.webp + pet.json from a single reference image, drop it into ${CODEXHOME:-$HOME/.codex}/pets/ / and Codex picks it up as a custom Codex Pet next to the 8 built-ins. This skill produces the exact Codex Pet atlas Codex expects (1536x1872 PNG/WebP, 8 cols x 9 rows, 192x208 cells, 9 animation states — idle, running-right, running-left, waving, jumping, failed, waiting, running, review). Calls OpenAI GPT Image 2 edit ONCE via the local RunComfy CLI as runcomfy run openai/gpt-image-2/edit to produce a canonical Codex Pet pose, then assembles all 9 animation rows programmatically with ImageMagick micro-transforms — no Codex Pro, no $imagegen, no OPENAIAPIKEY required, only RUNCOMFYTOKEN. Triggers on "codex pet", "create codex pet", "make codex pet", "hatch codex pet", "/hatch image", "desktop pet codex", "codex pets", "spritesheet.webp", or any explicit ask to build a custom pet for OpenAI Codex.
full-output-enforcement
leonxlnx/taste-skill
Overrides default LLM truncation behavior. Enforces complete code generation, bans placeholder patterns, and handles token-limit splits cleanly. Apply to any task requiring exhaustive, unabridged output.
executing-plans
obra/superpowers
Use when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available