unity-mcp

unity-mcp (MCP for Unity) is a Unity Editor extension that runs a local MCP/HTTP server on the user machine and exposes a rich set of “tools” and “resources” so AI assistants can inspect and modify Unity projects (assets, scenes, scripts, editor actions) and automate workflows.

Evaluated Mar 29, 2026 (23d ago)
Homepage ↗ Repo ↗ Ai Ml mcp unity editor-automation ai-integration local-server tooling csharp
⚙ Agent Friendliness
66
/ 100
Can an agent use this?
🔒 Security
44
/ 100
Is it safe for agents?
⚡ Reliability
48
/ 100
Does it work consistently?

Score Breakdown

⚙ Agent Friendliness

MCP Quality
85
Documentation
80
Error Messages
0
Auth Simplicity
90
Rate Limits
20

🔒 Security

TLS Enforcement
70
Auth Strength
30
Scope Granularity
10
Dep. Hygiene
40
Secret Handling
70

Network exposure is fail-closed by default: local loopback-only binding for HTTP local, explicit opt-in required for LAN binding and for plain HTTP remote. README mentions telemetry is anonymous and privacy-focused with an opt-out. However, no user authentication/authorization model (tokens/scopes) is described for the MCP server, so security relies heavily on local-only access and network settings. Dependency/version/CVE status is not stated in the provided content.

⚡ Reliability

Uptime/SLA
0
Version Stability
75
Breaking Changes
60
Error Recovery
55
AF Security Reliability

Best When

You want an AI-driven, local developer workflow where the Unity editor is already trusted and the AI is assisting with editor automation on the same machine.

Avoid When

You need multi-tenant or cross-user access, strong identity-based authorization, or fully unattended remote execution over untrusted networks.

Use Cases

  • Let an AI assistant create and edit Unity scenes (game objects, components, materials, textures)
  • Automate Unity editor tasks (builds, package management, menu actions, running tests)
  • Assist with script authoring and edits via code tools and optional Roslyn-based validation
  • Multi-instance Unity editor control by routing subsequent tool calls to a selected instance
  • Verifying Unity API usage by reflecting live C# APIs and fetching Unity documentation

Not For

  • Production remote control of Unity over the public internet without careful network hardening
  • Use cases requiring strong formal authentication/authorization between multiple users (it is local by default)
  • Environments where running arbitrary editor actions from an LLM tool layer is not acceptable

Interface

REST API
Yes
GraphQL
No
gRPC
No
MCP Server
Yes
SDK
No
Webhooks
No

Authentication

Methods: Local network only (loopback default) without user auth described in README
OAuth: No Scopes: No

README describes network defaults (fail-closed loopback by default; remote requires https and explicit opt-in for insecure/in-LAN binding) but does not describe user-level auth, tokens, or scopes.

Pricing

Free tier: Yes
Requires CC: No

No direct pricing for unity-mcp is stated; sponsorship and Coplay products are mentioned separately, but unity-mcp itself is MIT.

Agent Metadata

Pagination
none
Idempotent
False
Retry Guidance
Not documented

Known Gotchas

  • Statefulness: many tools change the Unity project/scene; repeated calls may have compounding effects
  • Long-running operations exist (async job tracking/polling); agent may need to poll using provided job endpoints rather than immediately retrying
  • Multi-instance routing requires selecting the active instance (set_active_instance) before subsequent actions

Alternatives

Full Evaluation Report

Comprehensive deep-dive: security analysis, reliability audit, agent experience review, cost modeling, competitive positioning, and improvement roadmap for unity-mcp.

AI-powered analysis · PDF + markdown · Delivered within 30 minutes

$99

Package Brief

Quick verdict, integration guide, cost projections, gotchas with workarounds, and alternatives comparison.

Delivered within 10 minutes

$3

Score Monitoring

Get alerted when this package's AF, security, or reliability scores change significantly. Stay ahead of regressions.

Continuous monitoring

$3/mo

Scores are editorial opinions as of 2026-03-29.

8642
Packages Evaluated
17761
Need Evaluation
586
Need Re-evaluation
Community Powered