Labsco
MCP SERVER

Roblox MCP

by hope1026

Let an AI build inside Roblox Studio — create and edit instances, scripts, UI, terrain and lighting live — across up to five Places at once, with changes reflected immediately.

3D, CAD & Game EnginesVerified
Summary
The AI can finally see inside Studio, not just hand you code.

The bridge is the whole point: without it an agent can only generate snippets you paste in, but WEPPY lets it create instances, edit scripts and shape terrain live, with the result visible in Studio immediately. The multi-Place support is the distinctive part — open several Studio windows for Lobby, Shop and Game and tell the agent which Studio ID to target, so one request can update Places that would otherwise each need context re-explained. Note much of the surface is PRO-gated.

What it is

The WEPPY MCP server, which connects an AI agent to the WEPPY Roblox Studio plugin so it can read and modify the DataModel directly: 24 tools spanning instances, scripts, UI, environment, assets, playtest control and arbitrary Luau execution. Many tools are marked [PRO].

What you get
  • The scene graph read and written: query_instances gets, finds and waits for instances, mutate_instances creates, deletes, clones, moves, renames and pivots them, and manage_properties gets and sets properties, attributes and tags
  • manage_scripts works script source directly — read, write, create, delete and edit lines — and execute_luau runs arbitrary Luau in the Studio sandbox (with unsafe services blocked)
  • manage_ui builds and verifies in-game UI (ScreenGui, Frame, TextLabel), and manage_lighting configures lighting, atmosphere, sky and terrain properties
  • Environment and motion: manage_tween for smooth animation, manage_audio, manage_animation, manage_physics collision groups, manage_effects for particles, and manage_terrain for fill, clear and material operations
  • spatial_query does raycasts, ground-finding and placement checks, and manage_camera and manage_selection control the view and what is selected
  • Assets: manage_assets inserts models by ID and searches the creator store, and manage_open_cloud_assets uploads through Roblox Open Cloud and returns an asset URI to apply
  • Workflow: manage_studio controls the playtest lifecycle and automated test runs, workspace_state snapshots and syncs the workspace, manage_sync handles project sync, and batch_execute runs several commands in one call
  • manage_logs reads and polls filtered output logs, and system_info reports connection status and usage tier
Requirements

Roblox Studio with the WEPPY Studio plugin installed — the server talks to the plugin, which is what actually touches Studio. Install via the one-line script from weppyai.com/en/install (install.sh on macOS/Linux, install.ps1 on Windows), which also sets up the client's agent plugin where supported. Asset upload uses Roblox Open Cloud credentials. Many tools are gated behind a PRO tier; system_info reports which tier you are on.

Setup effort

One command — npx -y @weppy/roblox-mcp@latest