Roblox Super MCP
Roblox Super MCP is a Node.js MCP server that connects MCP clients to Roblox Studio for project analysis, auditing, and safe, reversible script editing.
Discover and snapshot project structure: instances, scripts, remotes, modules, and live tree/place info.
Inspect exact Studio paths for properties, attributes, children, and script source.
Search live source code by literal text, Luau patterns, symbols, or method calls.
Read script source windows with hashing and ranked context.
Run dependency, data-flow, remote-contract, and race-condition analyses.
Analyze a system or script for relevance to a task before editing.
Audit scripts for lifecycle, performance, data integrity, typing, and safety issues.
Generate non-mutating change plans for features or fixes.
Apply atomic script edits with preconditions, dry-run, idempotency, and verification.
Create new instances/folders with validation and dry-run preview.
Verify recorded transactions against live Studio state and index drift.
Roll back changes with drift preconditions and optional dry-run or force.
Provides tools for interacting with Roblox Studio, including discovery, inspection, auditing, atomic script edits, verification, and rollback-oriented capabilities for Roblox projects.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Roblox Super MCP list all scripts in the current place"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Roblox Super MCP
Roblox Super MCP is a Node.js MCP server that connects an MCP client to the official Roblox Studio MCP process. It adds discovery, inspection, auditing, atomic script edits, verification, and rollback-oriented tools.
AI agents using the server should read and follow MCP_WORKFLOW.md. It defines the required private snapshot, diff-only changelog, change-note, and visual-version workflow.
Requirements
Windows
Roblox Studio with the official
StudioMCP.exeNode.js 18 or newer
An MCP-compatible client
Related MCP server: Roblox Studio Ultimate MCP Server
Install and test
From the project directory, run:
npm install
npm test
node server.jsThe server communicates over standard input and output. Stop the manual test
with Ctrl+C after confirming it starts successfully.
Replace an existing MCP launcher
If your current setup starts the original MCP through an mcp.bat file:
Back up the existing launcher.
Replace its contents with:
@echo off cd /d "<path-to-this-project>" node server.jsReplace
<path-to-this-project>with the folder where you cloned this repository. Keep the quotes if the path contains spaces.Restart the MCP client and Roblox Studio connection.
Do not print diagnostic messages to standard output. MCP messages use that stream; diagnostics should go to standard error or a log file.
Configure an MCP client directly
You can configure the client without a batch file. Add this server entry and replace the example path with the location of your local clone:
{
"mcpServers": {
"roblox-super-mcp": {
"command": "node",
"args": ["<path-to-this-project>/server.js"]
}
}
}Use forward slashes in JSON paths, or escape Windows backslashes as \\.
How it works
When started, the server locates the newest Roblox StudioMCP.exe in the
current Windows user's Roblox installation, starts it, and forwards validated
MCP requests to it. The available tool definitions are registered in
tool_registry.js; implementation helpers live in the tools directory.
Updating
Pull the latest version, test it, and restart the MCP client:
git pull
npm install
npm testKeep a backup of the previous launcher until the replacement has been tested in Roblox Studio.
Available Tools
23 toolsanalyze_projectA
(RECOMMENDED ENTRY POINT) Rebuilds the versioned project index from live Studio evidence and safely retains the last valid cache on failure.
Inputs: no named parameters.
apply_script_patchB
Atomically applies exact-text script edits with source-hash and match-count preconditions, dry-run, idempotency, verification, and rollback.
Inputs: path, edits, changes, dry_run, +2 more.
audit_data_integrityB
Audits ProfileStore/ProfileService/DataStore and receipt flows for session, pcall, persistence, yield, and grant-order safety.
Inputs: side, query, offset, max_scripts.
audit_lifecycleA
Audits live scripts for unowned connections, uncancellable threads, unbounded yields, stale references, and missing cleanup.
Inputs: side, query, offset, max_scripts.
audit_performanceA
Audits live hot paths for per-frame allocation, cloning, hierarchy scans, raycast pressure, polling, and missing pre-allocation.
Inputs: side, query, offset, max_scripts.
create_architectureA
Idempotently creates or updates folders, scripts, remotes, attributes, tags, and scalar/typed properties; created instances retain a transaction marker for safe rollback identity.
Inputs: nodes (required), dry_run, idempotency_key, create_missing_folders.
detect_race_conditionsB
Finds read-yield-use hazards, re-entrant handlers, concurrent state writes, and competing anchor authorities.
Inputs: side, query, offset, max_scripts.
find_dead_codeA
Returns confidence-scored unused module, function, remote, and attribute candidates without authorizing deletion.
Inputs: kind, limit, offset.
find_dependenciesB
Returns direct/transitive dependency and dependent graphs with cycles, unresolved requirements, ambiguity, and service usage.
Inputs: script_name (required), max_depth.
find_systemA
(RECOMMENDED) Ranked system discovery with match evidence across paths, functions, attributes, dependencies, services, and remotes.
Inputs: system_name (required), limit, offset, categories.
generate_change_planA
Produces a non-mutating, evidence-backed implementation and verification plan for a requested feature or change.
Inputs: request (required), limit, target, max_depth.
get_project_snapshotC
Returns a bounded project overview with place identity, counts, remotes, modules, cache provenance, and paginated indexed instances.
Inputs: limit, query, offset, refresh, +3 more.
get_remote_registryB
Lists remotes, callers/listeners, execution sides, methods, missing peers, naming findings, and evidence coverage.
Inputs: limit, offset, remote_name, include_unused.
impact_analysisA
(RECOMMENDED BEFORE EDITING) Calculates change risk from direct/transitive dependents, persistence, remotes, shared state, and ambiguity.
Inputs: change (required), max_depth.
inspect_instanceA
Batch-inspects exact Studio paths, returning properties, attributes, tags, children, and optional bounded script source.
Inputs: path, paths, properties, max_children, +2 more.
read_script_contextA
Reads a line-numbered live source window plus dependencies, dependents, remotes, functions, attributes, and source hash.
Inputs: path, line_count, start_line, script_name.
rollback_changeA
Reverts a recorded transaction in reverse order with drift preconditions and optional dry-run.
Inputs: transaction_id (required), force, dry_run.
sanity_check_scriptC
Performs a line-aware Luau audit for strict typing, lifecycle, remote validation, persistence safety, forbidden APIs, and modern standards.
Inputs: script_content (required), script_name.
search_sourceA
Searches live Luau source by literal text, Luau pattern, exact symbol frontier, or method call with line-numbered snippets.
Inputs: query (required), mode, limit, regex, +6 more.
trace_data_flowA
Traces observed remote and persistence participants for a state name while clearly separating observation from inferred flow.
Inputs: variable_name (required), limit, include_related.
validate_remote_contractA
Audits remote contracts for runtime validation, rate limiting, server authority, naming, missing peers, and client-supplied numbers.
Inputs: limit, offset, max_scripts, remote_name.
verify_changeA
Verifies a recorded transaction against live Studio state, reruns source sanity audits, and reports index drift.
Inputs: transaction_id (required).
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 23 tools
Most tools have distinct purposes, but find_system and get_related_code both rank files by match evidence across similar dimensions, which could cause confusion. The audit_* tools are differentiated by their target concern, so overall boundaries are mostly clear.
All tool names follow a consistent snake_case verb_noun pattern (e.g., analyze_project, get_project_snapshot, audit_lifecycle), making the set highly predictable and easy to navigate. There are no mixed conventions or vague names.
At 23 tools, the server is in the borderline heavy range, and the presence of near-duplicate tools like find_system and get_related_code suggests it could be slightly consolidated. The breadth of Roblox development is large, but the count feels marginally bloated rather than well-scoped.
The tool surface covers analysis, auditing, modification, verification, and rollback comprehensively for Roblox code. However, it lacks explicit instance deletion or standalone script creation (only via create_architecture), which are minor gaps agents could work around.
Maintenance
Related MCP Connectors
Remote MCP server exposing your Node Command fleet (files, shell, processes, browser) to AI hosts.
A MCP server built for developers enabling Git based project management with project and personal…
The official MCP Server for the Mux API
MCP server for understanding Javascript internals from ECMAScript specification.
Related MCP Servers
- AlicenseNot gradedqualityDmaintenanceAn agentic MCP server for Roblox Studio that provides over 45 tools for script diffing, playtest automation, and bulk instance manipulation. It enables AI agents to manage full development cycles within Roblox, including transaction-safe batch operations and real-time script reviews.2 npm1MIT
- AlicenseNot gradedqualityNot gradedmaintenanceA comprehensive MCP server providing 41 tools for complete Roblox Studio control, enabling safe script editing with syntax checking, instance manipulation, and backup features directly from Claude Code or any MCP-compatible AI assistant.1MIT
- AlicenseAqualityBmaintenanceDebug and inspection tooling for Roblox projects, exposed as an MCP server so an AI agent can explore the instance tree, read/write properties, call remotes, and run Luau in a Roblox client driven by an executor.1414 npmMIT
- AlicenseAqualityDmaintenanceDebug and inspection tooling for Roblox projects, exposed as an MCP server so an AI agent can explore the instance tree, read/write properties, call remotes, and run Luau in a Roblox client driven by an executor.1414 npm1MIT