MCP server for testing, debugging, and controlling Roblox Studio from AI coding tools
38 tools with highly inconsistent quality. Most tools have basic descriptions (10-50 chars) that lack actionable context for LLM selection. Parameter schemas are present but sparse: many lack type constraints, enums, or format specifications. No documented output schemas visible. Critical gaps: descriptions do not explain WHEN to use each tool vs. similar ones (e.g., set_properties vs. set_script_source), no error recovery guidance, no idempotency markers on write operations, and no permission/scope declarations. Tool names follow verb_noun convention (positive), but parameter naming is inconsistent (e.g., 'instance_id' appears in 30+ tools but is never explained). Write-heavy tool set (15+ WRITE tools) with no dry-run or confirmation patterns. Baseline: 549 production tools average 194-char descriptions; this server averages ~40 chars.
Manage breakpoints in the debugger
Capture micro profiler data
Capture a screenshot of the Studio viewport
Capture script profiler data
Delete lines from a script
Edit specific lines in a script
Evaluate code in the client runtime
Descriptions critically short (avg 30 chars vs. baseline 194 chars). Most lack WHEN/WHY context. E.g., 'Set properties on an instance' does not explain when to use set_properties vs. set_script_source or edit_script_lines. LLMs cannot disambiguate.
No output schemas documented. Tools return data but LLMs cannot plan downstream calls or extract required fields. E.g., get_instance_properties returns what fields? Does it include child references needed for chaining?
15+ WRITE tools (set_properties, set_script_source, edit_script_lines, delete_script_lines, execute_luau, etc.) lack idempotency markers, dry-run options, or confirmation patterns. Agents cannot safely retry on ambiguous failures without risking duplicate side effects.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 44 | 2026-07-28+ | v2 |
Evaluate code in the server runtime
Execute Luau code in Studio
Export instances as an RBXM file
Find and replace text in scripts
Generate a model from a description
Get attributes of an instance
Get list of connected Studio instances
Get properties of a specific instance
Get memory usage breakdown
Get information about the current place in Studio
Get the hierarchical structure of the project
Get the status of a pending request by request ID
Get Roblox API documentation
Get Roblox-related skills and documentation
Get runtime logs from the Studio session
Get analysis of the current scene
Get the source code of a script
Search for patterns in scripts
Import instances from an RBXM file
Insert an asset into the place
Insert new lines into a script
Manage Studio instance lifecycle (launch, authorize, complete, close, status)
Manage multiplayer playtest sessions
Preview an asset before inserting
Search for objects in the DataModel
Get or set the current selection in Studio
Set properties on an instance
Set the source code of a script
Simulate keyboard input in Studio
Simulate mouse input in Studio
Start or stop a solo playtest session
Parameter descriptions missing or trivial. 'instance_id' appears in 30+ tools but is never explained (is it a UUID? hostname? process ID?). 'action' parameter used in 5+ tools with no enum or valid values documented.
No error recovery guidance. Tools do not document what errors are retryable, what requires user intervention, or what alternatives exist. E.g., if search_objects returns empty, should the LLM try a broader query or a different search type?