Safeguard Desktop App Guide: Windows, macOS and Linux

The Safeguard desktop app (version 3.2.2) runs on Windows 10 and 11, macOS 12 and later, and x86_64 Linux. It puts the full Safeguard web app in a native window and adds a local panel that scans folders you grant and runs tasks you approve. You ask your supply chain questions in plain English.

It is also the easiest place to see what switching to Safeguard looks like without giving anything up. Bring the Claude, ChatGPT or Codex plan you already pay for and use it inside the Safeguard CLI. Safeguard adds no extra charge for that usage; your plan's own usage limits and terms still apply. And if you already work in Claude or ChatGPT, connect either one to the Safeguard MCP server and use Safeguard from the chat you already use.

This guide covers each system in turn: which file to download, how to check it, what the first launch looks like, what to ask, and how the desktop app fits with the CLI in your terminal and the app on your phone. Device facts come from the Safeguard download page as checked on 21 October 2026.

The desktop app at a glance

The desktop documentation describes the app as a wrapper around the web app, not a replacement for it. The window has two parts:

• The web app on the left, about three quarters of the width. It is the same platform you would open in a browser, and you sign in exactly as you would there.
• A local panel on the right, about a quarter of the width. It can reach folders you grant, a restricted terminal and MCP servers. Toggle it with Ctrl+Shift+S on Windows and Linux or Cmd+Shift+S on macOS, or from the View menu.

That split is the point. The left side answers questions about your whole estate. The right side works on the machine in front of you.

The Windows installer works per user and needs no admin rights. Every release also publishes a SHA256SUMS file with the full checksum for each desktop build.

Install and verify on each system

Verifying a download takes one command and tells you the file you have is the file Safeguard published. Do it on every system, and especially on Windows while the builds are unsigned.

Windows 10 and 11

1. Download Safeguard-3.2.2.exe from https://safeguard.sh/download. It picks x64 or ARM for you. If you want a smaller file, take the architecture-specific installer instead.
2. Check the SHA-256 before you run it. The download page shows the expected value; for the 3.2.2 universal installer it begins with 069f5fab80cd100e.

certutil -hashfile "$env:USERPROFILE\Downloads\Safeguard-3.2.2.exe" SHA256

# or in PowerShell, where SHA256 is the default algorithm
Get-FileHash "$env:USERPROFILE\Downloads\Safeguard-3.2.2.exe"

3. Run the installer. Because the builds are not yet code-signed, Microsoft Defender SmartScreen may show a warning. Microsoft's documentation explains why: SmartScreen checks downloads against a list of files that are well known and frequently downloaded, and warns when a file is not on it. A new, unsigned release has no reputation yet. Once the checksum matches, choose More info, then Run anyway.
4. The app installs for your user only. No administrator prompt is expected.

One note for Windows on ARM laptops: the desktop installer covers ARM natively. The Safeguard CLI has no native Windows ARM build; the CLI docs say the PowerShell installer warns and installs the x86_64 binary, which runs under emulation.

macOS 12 and later

1. Check your chip with Apple menu, About This Mac. Apple Silicon takes Safeguard-3.2.2-arm64.dmg; Intel Macs take the Intel build linked on the same page.
2. Verify the disk image. The Apple Silicon checksum begins with cb5ac4f57a5ff6e6 and the Intel one with 402cdfa6473dc2f2.

shasum -a 256 ~/Downloads/Safeguard-3.2.2-arm64.dmg

3. Open the DMG and drag the app to Applications.
4. Optionally confirm Gatekeeper's view. The Safeguard desktop docs say spctl should report "accepted, source=Notarized Developer ID".

spctl --assess --type execute -vv /Applications/Safeguard.app

Apple explains that for apps distributed outside the App Store, macOS checks the Developer ID signature and that notarization means Apple checked the software for malicious components without finding any. That is why the Mac builds open without the warnings you might see for unsigned software.

Linux (x86_64)

1. Confirm the machine with uname -m; it should print x86_64. There is no arm64 desktop build.
2. Install the Safeguard CLI first. The download page says the Linux desktop build also needs the CLI installed and on your PATH.
3. Download the AppImage and verify it. The 3.2.2 checksum begins with 46685975126a0d01.

curl -fsSL https://cli.safeguard.sh/install | bash
which safeguard

cd ~/Downloads
sha256sum Safeguard-3.2.2-x86_64.AppImage
chmod +x Safeguard-3.2.2-x86_64.AppImage && ./Safeguard-3.2.2-x86_64.AppImage

If it stops with "AppImages require FUSE to run," install your distribution's FUSE 2 package. The AppImage project lists libfuse2 for Ubuntu 22.04 and later, libfuse2t64 for Ubuntu 24.04, and fuse2 on Arch.

Updates and uninstalling

The desktop docs say the app checks for updates on launch and every four hours. In Settings you choose one of three modes:

• Auto-install installs the update when you quit.
• Ask before install notifies you and waits. This is the default.
• Manual checks only when you click Check for Updates.

To uninstall, drag the app to the Trash on macOS, use Settings, Apps, Installed apps on Windows, or delete the AppImage on Linux.

First run and what you can ask

Sign in, then open the panel

Open the app and sign in in the left pane, the same way you would in a browser. Then press Ctrl+Shift+S (Cmd+Shift+S on a Mac). The panel opens on the Files tab, where you grant folders through your operating system's file picker. The app can only read folders you have granted, and you can revoke them later from the same tab.

The six panel tabs

The desktop docs describe six tabs:

• Scan. Add local projects and run quick or full vulnerability scans. Results group into critical, high, medium, low and unknown. Auto-scan can be switched on per project.
• Terminal. Runs an allowlist of binaries only, including safeguard-cli, git, docker, npm, python and curl, with the working folder inside a granted folder.
• MCP. Start and stop MCP servers, including a one-click Safeguard server, or connect external servers over HTTP or SSE. Each shows starting, running, stopped or crashed.
• Tasks. Describe a job in plain language. You approve or deny destructive actions before they run, cancel running tasks and schedule recurring ones.
• Files. Grant and revoke folder access and browse what you granted.
• Settings. Update mode, theme, minimize to tray, the action approval mode, prompt injection detection, and version and platform details.

What to ask

There are two places to ask, and they answer different kinds of question.

In the web app on the left, ask about your whole supply chain in plain English, the same as you would in a browser. For example:

• "How many critical vulns do we have?"
• "Which projects have the most critical findings?"

In the Tasks tab on the right, ask for work on this machine. The docs say Tasks matches requests to built-in workflows for a local repo scan, an MCP configuration audit, a dependency watch, a container scan and an IDE extension check, and breaks anything else into steps. For example:

• "Scan the repos in my granted folders and list anything critical."
• "Audit my MCP configuration."
• "Check my IDE extensions."
• "Scan the image I just built." (Container scans need a running Docker or Podman daemon.)
• "Re-scan my granted folders every 24h and flag new critical findings."

Recurring tasks

Recurring intervals use a number and a unit: 30s, 5m, 1h or 24h. Cron expressions are not accepted, and the app reports "Invalid interval format" if you try one.

Shortcuts worth learning

Ctrl or Cmd plus N starts a new scan, plus O opens a project, and plus comma opens settings. The app also registers safeguard:// links that open a panel tab such as scan or tasks, or send the web view to dashboard.

The web view only navigates within safeguard.sh domains. External links open in your normal browser.

How it pairs with the CLI, your phone and Claude or ChatGPT

Safeguard is one platform on every surface, so the desktop app is one stop in a longer day rather than a separate product.

Start in the terminal. The CLI scans, explains and fixes in the folder you are working in. The desktop docs say the app's own scanner uses a safeguard-cli binary too, and on Linux the download page asks you to have the CLI on your PATH.

Continue on the desktop. The left pane shows the same tenant you scanned into from the terminal. The right panel keeps watching your granted folders on a schedule.

Check on your phone. Android phones and tablets install from Google Play. A direct APK (1.0.2) with a published SHA-256 is offered for phones without Google Play or for company-managed rollout. iPhone and iPad install from the App Store, and the Safari extension on iPhone and iPad comes inside the iOS app; turn it on in Settings, Apps, Safari, Extensions.

Use your browser. The browser extension (1.0.4) is on the Chrome Web Store for Chrome, Brave, Arc, Vivaldi, Opera and Edge, and on Firefox Add-ons. Safari on Mac is in Apple review and not yet available.

Or install nothing. The web interface is the full platform in a browser. The Cloud CLI and cloud computer are hosted by Safeguard, and the cloud sandbox gives you an isolated tenant against a seeded demo repo, free for seven days with no card.

Keep Claude or ChatGPT. If your team already lives in a chat app, connect it to the Safeguard MCP server. Anthropic's help centre says custom connectors are available on Claude's Free, Pro, Max, Team and Enterprise plans, with Free limited to one, and that most connectors sign in with OAuth so you never hand Claude a password. This demo shows the Claude side: connect Claude to the Safeguard MCP server with OAuth and choose which tools Claude can use.

Click through it at your own pace, or open it full screen.

The same flow exists for ChatGPT. If you are weighing chat assistants for coding generally, see Claude vs ChatGPT for coding.

Do the work properly with the Safeguard CLI

The desktop app is where you look and ask. The CLI is where you do the work. It scans code, container images and SBOMs on your own machine, sends only the results to the platform, and includes an assistant that can fix what it finds. It writes code, not only scans.

Install

# macOS and Linux
curl -fsSL https://cli.safeguard.sh/install | bash

# Windows
irm https://cli.safeguard.sh/install.ps1 | iex

It is a single static binary, with no Node, Python or JVM needed. It runs on macOS, Linux and Windows, on x86_64 or arm64 (Windows is x86_64 only). The installer verifies the download and puts safeguard on your PATH. Confirm with safeguard --version.

First session

Run safeguard with no arguments to open the interactive terminal. /login signs you in through the browser, and /scan . scans the current folder. Then ask in plain language: "how many critical vulns do we have" or "fix the lodash CVE in this repo." Type @ to attach a file from the folder you started in, and /resume picks up the last session in that folder.

The assistant has tools for your local files and for your Safeguard data and picks whichever the question needs. So it is a general terminal assistant as well as a security one. "What's the latest file in ~/Downloads" is one of the examples in the docs. With your approval it can find and read files, edit code and run commands.

Writing and fixing code

Ask for the change you want. The assistant proposes edits and asks before writing them. Because it also knows your vulnerabilities, a fix request is grounded in what is actually flagged for that project. When another agent did the writing, scan the folder before you merge; this walkthrough shows how.

This demo shows the same work started from Search: attach a computer running the Safeguard CLI and a folder, ask for a fix, review the diff and approve the command before it runs. The demo notes that computer work from Search is available and expanding.

Click through it at your own pace, or open it full screen.

Use your Claude or ChatGPT plan

Bring the Claude, ChatGPT or Codex plan you already pay for and use it inside the Safeguard CLI. Safeguard adds no extra charge for that usage; your plan's own usage limits and terms still apply. To connect a model, see the CLI docs or type /model, which shows or switches the AI model.

Approvals

Tools that change things, such as writing a file, editing code or running a command, ask before they act. /permissions sets the mode: Ask prompts before every action, Auto approves known actions and asks about anything new (the default), and Yolo approves everything. The desktop app has its own action approval mode in Settings, so set both to match how careful you want to be.

CI with safeguard exec

safeguard scan --dir .
safeguard validate ./sbom.cdx.json
safeguard exec "summarise the critical findings in this repository"

safeguard scan also takes --repo, --image or --sbom, and writes JSON results such as vuln.json and license.json under ./output. safeguard validate checks CycloneDX, SPDX and VEX documents. safeguard exec runs one non-interactive assistant turn, which suits CI. When the local runner runs platform scan jobs, your source never leaves the machine; only SBOMs and findings are uploaded. Policies and gates are evaluated in the platform, not by the CLI. For a longer tour, read the Safeguard CLI as a security-aware coding assistant.

Common mistakes

• Clicking Run anyway before checking the hash. SmartScreen warns because the Windows builds are unsigned and new. The checksum is what tells you the file is genuine. Check it first.
• Downloading the Apple Silicon DMG for an Intel Mac. Look at About This Mac and take the Intel build if that is what you have.
• Expecting a Linux desktop build on arm64. There is only x86_64. Use the CLI and the web interface.
• Launching the AppImage before the CLI is installed. The download page says the Linux build needs the CLI on your PATH.
• Forgetting to grant a folder. The panel cannot see anything you have not granted on the Files tab. If a Terminal command is rejected, check the folder and the allowlist.
• Asking the Tasks tab platform-wide questions. Questions about your whole estate belong in the web app on the left. The Tasks tab is for work on this machine.
• Writing a cron expression for a recurring task. Use 24h, not 0 9 * * *.
• Container scans with no daemon running. Start Docker or Podman first.
• Looking for the app in the Microsoft Store or Galaxy Store. It is not listed there. Use the download page.

FAQ

Which systems does the Safeguard desktop app support?

Windows 10 and 11 on x64 and ARM, macOS 12 and later on Apple Silicon and Intel, and Linux on x86_64 as an AppImage. The current version on the download page is 3.2.2.

Why does Windows warn me when I run the installer?

The Windows builds are not yet code-signed, so SmartScreen has no reputation for them and may warn. Verify the SHA-256 with certutil -hashfile or Get-FileHash, then choose More info, then Run anyway.

Do I need admin rights on Windows?

No. The installer installs per user without admin rights.

Can I switch to Safeguard and keep my Claude, ChatGPT or Codex plan?

Yes. Bring the Claude, ChatGPT or Codex plan you already pay for and use it inside the Safeguard CLI. Safeguard adds no extra charge for that usage; your plan's own usage limits and terms still apply. You can also connect Claude or ChatGPT to the Safeguard MCP server and use Safeguard from those apps.

What is the difference between the desktop app and the web app?

The left side of the desktop app is the web app. The desktop app adds the right panel, which can scan folders you grant, run allowlisted commands, manage MCP servers and run scheduled tasks on your machine.

Does the desktop app act without asking?

Destructive actions in the Tasks tab ask for approval before they run, and you can cancel a task while it runs. The action approval mode is in Settings.

Is there a phone app?

Yes. Android from Google Play, or a direct APK (1.0.2) for phones without Google Play, and iOS from the App Store for iPhone and iPad.

Can I try it without installing anything?

Yes. The web interface needs nothing installed, and the cloud sandbox is an isolated tenant against a seeded demo repo, free for seven days with no card.

Try it yourself

This demo is the CLI half of the setup: set up the runner on your own machine, pick the platform, start it, approve it with its device code, then use the same CLI to scan from the terminal.

Click through it at your own pace, or open it full screen.

Get the desktop app, the CLI and the phone apps from https://safeguard.sh/download, and keep https://docs.safeguard.sh/docs/cli open for the commands.

Quick check

Sources

• Safeguard downloads
• Safeguard desktop application docs
• Safeguard CLI overview
• Safeguard CLI installation
• Safeguard CLI interactive terminal
• Safeguard CLI slash commands
• Microsoft Defender SmartScreen overview
• certutil reference
• Get-FileHash reference
• Apple: Safely open apps on your Mac
• Claude: custom connectors using remote MCP
• AppImage FUSE troubleshooting

Last reviewed: 21 October 2026.