Installation

This page provides a detailed walkthrough of the NexusFlow installation process. NexusFlow consists of two components that need to be installed separately:

• Desktop App — Free, downloaded from GitHub
• UE Plugin — Purchased from Fab.com and installed into your UE engine

Prerequisites

Item	Minimum Requirement
Operating System	Windows 10 / 11 (64-bit)
Unreal Engine	Unreal Engine 5.6 or 5.7
LLM API Key	OpenAI / Anthropic / Google Gemini / DeepSeek / Zhipu GLM / MiniMax / Qwen / Kimi supported
Network	Internet connection required for AI model API calls

UE Version Compatibility

NexusFlow requires UE 5.6 or 5.7. UE 5.5 and earlier are not supported because they lack ONNX Runtime 1.20+, which is needed for the RAG embedding model.

Install the Desktop App (from GitHub)

The NexusFlow desktop app is completely free and always will be.

Download

Go to the GitHub Releases page and download the latest .exe installer.

Install

1. Double-click the downloaded .exe installer (NSIS installer)
2. Select your preferred language (Simplified Chinese / English)
3. Choose whether to install for the current user or all users
4. Follow the setup wizard to choose an installation directory
5. Complete the installation

After installation, launch NexusFlow from the Start Menu or desktop shortcut. On first launch, a floating ball icon will appear on your desktop and the first-run setup wizard will guide you through configuration.

Install the UE Plugin (from Fab)

NexusFlow requires a companion plugin installed in the UE Editor to communicate with the engine. The plugin is available on Fab.com.

Option 1: Install from Fab (Recommended)

1. Purchase the NexusFlow plugin on Fab.com
2. In the Epic Games Launcher, go to your Fab Library
3. Find NexusFlow and click Install to Engine
4. Select UE 5.6 or 5.7 as the target engine version
5. Restart the UE Editor

Option 2: Manual Installation

If you have the plugin as a zip file:

1. Close the UE Editor
2. Extract the NexusFlow folder into your UE project's Plugins/ directory
3. Reopen the UE Editor
4. Go to Edit > Plugins, search for "NexusFlow", and confirm it is enabled

Verify Plugin Installation

You can check the plugin installation status from NexusFlow's Settings > About page:

• ✅ Plugin detected — Installation successful
• ❌ Plugin not found — Review the installation steps or reinstall

Click the "Re-detect" button to manually refresh the detection result.

UE Plugin First Launch

When the UE Editor starts with the NexusFlow plugin installed, the plugin automatically checks your environment. If anything is missing, a NexusFlow Setup dialog appears.

What Gets Checked

Item	What It Checks	How
NexusFlow Desktop App	Whether the app is installed on this machine	Windows Registry lookup
Embedding Model	Whether the RAG search model is ready	Bridge FFI status check

When the Dialog Appears

• Everything ready — No dialog; the plugin starts silently
• Something missing — The setup dialog appears after a short delay (~2 seconds)
• "Don't Show Again" was selected — No dialog, even if items are missing

Dialog Options

Button	Effect
Download Model	Starts downloading the embedding model (~90 MB) in the background
Open Settings	Opens Project Settings > Plugins > NexusFlow for manual configuration
Later	Dismisses the dialog; it will appear again on next startup
Don't Show Again	Permanently hides the dialog (can be re-enabled in settings)

Not-Installed App Links

If the NexusFlow desktop app is not detected, the dialog provides two download links:

• GitHub — Releases page
• Mirror — Accelerated download link for users in China

Re-enabling the Setup Dialog

If you previously clicked "Don't Show Again" and want to see the dialog again:

1. Navigate to the NexusFlow config file: %LOCALAPPDATA%/NexusFlow/config.json
2. Set "showSetupWizard": true
3. Restart the UE Editor

Alternatively, you can check environment status at any time in Project Settings > Plugins > NexusFlow.

Configure AI Models

NexusFlow requires at least one AI model to be configured before use. There are two ways to set this up:

Option 1: First-Run Wizard

On first launch, Step 2 of the wizard will guide you through AI model configuration. See Quick Start for details.

Option 2: Settings Page

1. Click the gear icon at the top of the sidebar to open Settings
2. Select AI Models
3. Click "Add Model"
4. Fill in the configuration fields (see below)

Supported AI Model Providers

Provider	Example Models	Notes
OpenAI	gpt-5.4, gpt-5.3-codex	OpenAI official API
Anthropic	claude-opus-4-6, claude-sonnet-4-20250514	Max Tokens is required
Google Gemini	gemini-3.1-pro-preview, gemini-2.5-pro	Google AI Studio API
DeepSeek	deepseek-chat, deepseek-reasoner	Cost-effective option
Zhipu GLM	GLM-4.7, GLM-5	Model names are case-sensitive
MiniMax	MiniMax-M2.7, MiniMax-M2.7-highspeed	Max Tokens is required
Qwen	qwen3-max, qwen3.5-plus	Alibaba Cloud DashScope API
Kimi	kimi-k2.5, kimi-k2-turbo-preview	Moonshot API
Custom	User-defined models	Supports any API compatible with the above protocols

Configuration Fields

Field	Required	Description
Provider	✅	Select the model provider
Model Name	✅	Custom display name shown in the model list
API Key	✅	API key provided by the provider
Model ID	✅	Model identifier, e.g., gpt-5.4
API URL	Required for Custom	Custom API endpoint URL
Protocol	Required for Custom	OpenAI Compatible / OpenAI Responses API / Anthropic Messages API / Google Gemini API
Temperature	Optional	Generation randomness (0–2); higher values produce more varied outputs
Max Tokens	Optional*	Maximum output tokens. *Required when using the Anthropic protocol

TIP

You can add multiple models and set one as the default. Click a model card in the list to view its details and manage it.

API Key Security

API Keys are displayed in masked form (sk-****) in the UI. Click the eye icon to toggle visibility. API Keys are stored locally and are never uploaded to any server.

Verify Your Setup

After installing both components, run through this checklist to confirm everything is working:

1. Launch NexusFlow — The floating ball appears on screen
2. Click the floating ball — The sidebar opens normally
3. Check models — Settings > AI Models shows at least one model with a passing validation status
4. Launch UE — The sidebar status bar shows a green connection indicator
5. Send a message — Try typing "Hello" to confirm the AI responds

If you encounter any issues, see Troubleshooting.