AI coding assistants like Claude are usually tied to a laptop or workstation. But for many developers, there’s a new need: an AI that can work autonomously, without the computer being turned on, and with direct access to projects stored on a NAS.

if you have Synology NAS and is only use for storing your documents let’s make it as your AI assistant.

In this article, we’ll set up a Claude Agent on a Synology NAS using Container Manager, so the NAS becomes a 24/7 AI coding host. The model itself still runs in the cloud via Claude API; the NAS is responsible for hosting the agent, workflows, and automation.

Why Run AI on a NAS?

There are several practical reasons to host your AI agent on a NAS:

• Always-on device – NAS is typically running 24/7
• Projects live on NAS – the agent can read and write code directly
• No need to keep your PC on – automation continues even when your workstation is off
• Accessible from any device – phone, tablet, or laptop
• Supports automated workflows – refactoring, documentation, PR review, and more

With this setup, the NAS becomes an AI automation node, while Claude API remains the “brain” that performs reasoning and code generation.

If you have NAS more than 8 GB RAM, you can run your AI OLLAMA in local. but if you only have 4 GB this setup is work perfectly

Architecture Overview

Code

Claude API (cloud)
        ↓
Open WebUI (Agent Host) on Synology NAS
        ↓
Project folders / Git repos on NAS
        ↓
Browser (phone / laptop / tablet)

Open WebUI acts as the agent engine that runs coding workflows, and Claude API provides the intelligence for code understanding, generation, and reasoning.

Prerequisites

• Synology NAS with DSM 7.2+
• Container Manager installed
• A project folder on NAS (e.g., /volume1/git)
• A valid Claude API Key
• Stable internet connection

No SSH or terminal is required; everything can be done through the DSM GUI.

Step-by-Step: Installing Claude Agent on Synology NAS

1. Prepare a Folder for Open WebUI Data

In File Station, create:

Code

/volume1/docker/openwebui

This folder stores configuration and agent data.

2. Deploy Open WebUI via Container Manager

Open Container Manager → Create → Deploy Container

Image:

Code

ghcr.io/open-webui/open-webui:latest

Click Next.

3. Configure the Container

A. General Settings

• Name: openwebui
• Restart policy: Always

B. Port Settings

Access later via:

Code

http://nas.local:3000

C. Volume Settings

Add two mounts:

1. Open WebUI Data

• Host: /volume1/docker/openwebui
• Container: /app/backend/data

2. Project Folder

• Host: /volume1/git
• Container: /workspace

This allows the agent to read and write project files.

D. Environment Variables

Add:

E. Execution User

Set:

Code

root

This allows the agent to modify files in your project folders.

4. Deploy and Start the Container

Click Done → Apply → Start.

Open:

Code

http://nas.local:3000

You should see the Open WebUI interface.

Configuring Claude in Open WebUI

Go to:

Settings → Models → Add Model → OpenAI-Compatible API

Fill in:

• Base URL: https://api.anthropic.com
• API Key: your Claude API key
• Model: claude-3-5-sonnet or claude-3-5-haiku

Open WebUI will now use Claude API as its engine.

Enabling Agent Mode

Go to:

Settings → Agents → Enable Agent Mode

Agent Mode enables:

• Reading files from /workspace
• Writing and modifying files
• Automated refactoring
• Documentation generation
• Code review
• Running workflows without supervision

Your NAS is now a Claude Agent Host.

Can the Agent Read GitHub Repositories?

Yes — two methods:

1. Clone GitHub Repos to NAS (Most Stable)

Code

cd /volume1/git
git clone https://github.com/username/repo.git

The agent can now read the entire codebase locally.

2. Access GitHub via API

Add a GitHub Personal Access Token (PAT):

The agent can fetch files, inspect PRs, and combine GitHub data with local project context.

Example 24/7 Coding Workflows

Automated Refactoring

Code

Read the entire folder /workspace/projectA.
Create a refactoring plan.
Apply the refactor step by step.
Write the updated files directly into the folder.
Run until the refactor is complete without waiting for my input.

Documentation Generation

Code

Scan the entire folder /workspace/projectA.
Generate full documentation into /workspace/projectA/docs.
Include architecture overview, API endpoints, and sequence diagrams in Markdown.

Pull Request Review

Code

Review PR #12 from the repository.
Identify potential bugs and design issues.
Save the review into /workspace/projectA/review-pr12.md.

These workflows can be triggered manually or scheduled, turning your NAS into a continuous AI assistant.

Conclusion

By combining Synology’s Container Manager, Claude API, and Open WebUI, you can transform your NAS into a 24/7 AI coding assistant host. The NAS handles agents and workflows, while Claude provides the intelligence.

This setup is ideal for developers who store projects on a NAS and want automated refactoring, documentation, and code review without keeping a workstation powered on.