How to Connect Claude to the Safeguard MCP Server, Step by Step

To connect Claude to Safeguard, add Safeguard from Claude's connector directory, or add a custom connector with Safeguard's Claude address. Claude then opens Safeguard's sign-in page, you check the consent page and click Authorize access. No API key is needed. You decide which tools Claude can use, and you can revoke access any time.

This guide covers every route, then the consent page, tool access, revoking, and what to ask Claude.

Safeguard is also something you can switch to without losing what you already pay for. Keep using Claude through the Safeguard MCP server, and 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.

Ways to connect Claude, and what each one needs

Safeguard hosts its MCP server, so there is nothing to install on your machine. Safeguard's docs say plainly that it publishes no npm package or local bridge for the server, and that any package claiming to be the Safeguard MCP server should be treated as untrusted.

There are two addresses worth knowing. The general MCP endpoint is https://mcp.safeguard.sh/mcp, which is what the download page lists and what clients such as Claude Code, ChatGPT and Cursor use. Safeguard's Claude guide gives a Claude connector address, https://mcp.safeguard.sh/mcp/anthropic, for the custom connector form in claude.ai and Claude Desktop. The MCP Server card inside Safeguard, under Integrations › Deployment › MCP Server, shows the same Claude address.

From Anthropic's help center: custom connectors using remote MCP are available on Claude, Cowork and Claude Desktop for Free, Pro, Max, Team and Enterprise plans, and Free users are limited to one custom connector. Claude reaches remote MCP servers from Anthropic's cloud, not from your device, so the server must be reachable over the public internet. Safeguard's server is hosted publicly, so that requirement is already met.

Connect Claude to Safeguard, step by step

Before you start, create your Safeguard account at app.safeguard.sh or sign in to it. The connector acts as you, so use the account that has the access you intend Claude to have.

Route 1: the connector directory

Safeguard's download page says you can add Safeguard from Claude's connector directory on web, desktop or mobile, and sign in once with OAuth, with no API key to paste. There is no URL to type.

1. Sign in to claude.ai, then open Safeguard's entry in Claude's connector directory. Safeguard's Claude guide links to it directly; without a Claude session, the link lands on the Claude home page.
2. Select Safeguard, review its description, and connect.
3. Claude sends you to Safeguard's sign-in page. Sign in, check the consent page (more on that below), and click Authorize access.

Safeguard's docs add that if the directory is not available on your plan or in your workspace, the custom connector steps always work.

Route 2: a custom connector with Safeguard's Claude address

This is the manual route from Safeguard's Claude guide, and it matches the steps in Anthropic's custom connector article for Pro and Max.

1. In Claude, open Customize › Connectors, click + Add, then Add custom connector.
2. Enter the name Safeguard Security and the URL https://mcp.safeguard.sh/mcp/anthropic, then click Continue.
3. Claude detects that Safeguard signs in with OAuth. Click Continue.
4. Under Authentication, choose Sign in now or Sign in when needed.
5. Under OAuth client, choose Register automatically. Safeguard accepts only client IDs it issued itself, so do not pick an option that brings your own client.
6. Click Add.
7. When Claude sends you to Safeguard, sign in, check the app name and the host on the consent page, then click Authorize access.

Route 3: Team and Enterprise organizations

On Team and Enterprise plans, an Owner or Primary Owner adds the connector for the organization first.

1. Go to Organization settings › Connectors.
2. Click Add, then Custom, then Web.
3. Use the same name, Safeguard Security, the same URL, and the same Authentication and OAuth client settings as above.
4. Click Add.

Members then open Customize › Connectors, find Safeguard Security, and click Connect. Each member signs in to Safeguard individually and sees only their own access.

Claude Desktop and mobile

Claude Desktop uses the connectors in your Claude account, so once Safeguard Security is added it is there too. Safeguard's guide says not to add Safeguard to claude_desktop_config.json, which is for local servers only.

Anthropic's connectors article says services connected on web or Desktop appear on mobile at your next login, and that installing connectors from mobile is in beta. The simplest path is to connect on web or Desktop and then use Safeguard from the iOS or Android app.

Claude Code

Claude Code can add Safeguard as a remote HTTP server. Safeguard's connector setup page gives this command:

claude mcp add --transport http safeguard https://mcp.safeguard.sh/mcp

Then run /mcp inside Claude Code and follow the steps in your browser to sign in. Anthropic's Claude Code docs describe the rest of the lifecycle:

claude mcp list            # see configured servers
claude mcp get safeguard   # details for one server
claude mcp remove safeguard

Removing a remote server also deletes the OAuth tokens Claude Code stored for it, and Clear authentication in the /mcp menu revokes access without removing the server. The default scope is local, for you in the current project; add --scope user to use it across your projects.

There is a shortcut too. If you log in to Claude Code with a claude.ai account, Anthropic's docs say the connectors you added in claude.ai are automatically available in Claude Code. They are not loaded when an API key or a third party provider is the active authentication.

This demo connects Claude to the Safeguard MCP server with OAuth and lets you choose which tools Claude can use.

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

The consent page, choosing tools, and revoking access

What the consent page shows

Safeguard's consent page names the assistant, Claude, and the host that will receive the authorization code, claude.ai. Safeguard's demo explains that both values come from the MCP server's check of the signed request, not from the link you clicked, and that the code shown matches the one Claude shows.

If the page names Claude and claude.ai and you started the connection from Claude, click Authorize access. If it names a different app or host, stop. Anthropic's article gives the same advice: review the permissions requested during authentication and deny access you do not need.

How tool access is narrowed

Claude gets your own Safeguard access, never more. Several layers then narrow it, and Safeguard's demo puts the rule well: each step can only narrow the one before.

On the Safeguard side, a new workspace starts with six read-only tools: list products, list projects, get a project, list vulnerabilities, get one vulnerability, and list findings. Every one is a read. Everything else, including tools that start a scan or add an integration, is off until an administrator turns it on under Feature Flags. Safeguard's tool gating page says gating is per tenant, not per user, and that if tool resolution ever fails, the server falls back to the small default set rather than exposing anything wider.

Safeguard's demo shows the same idea through MCP access. With read tools on and write tools off, Claude can list projects, find vulnerabilities and read SBOMs, but it cannot start scans or open fix pull requests until an admin allows write tools.

On the Claude side, Anthropic's article says Claude asks for approval before it invokes tools, and that you should only click Allow always for servers and tools you trust to run unsupervised. You can turn the connector on or off per conversation from the + button, then Connectors. Use the Search and tools menu to disable tools you do not want Claude to use. One caution from the same article: in Research, Claude can invoke connector tools without further approval, so disable any write tools before you use Research with a connector.

Revoking access

You can end Claude's access from either side.

• In Safeguard: open Settings › MCP connections. It lists Claude as an AI connection, with when it was last used and when the grant expires. Revoke it there at any time.
• In Claude: open Customize › Connectors and remove or disconnect the connector. Team and Enterprise Owners can do this for the organization in Organization settings › Connectors.
• In Claude Code: use Clear authentication in /mcp, or claude mcp remove safeguard.
• For MCP keys: revoke the key under Settings › Developer. Safeguard's docs say revocation takes effect within a minute.

MCP keys exist for clients that can only send a fixed Authorization header. A key acts as the member who created it, across the whole workspace, so keep it out of source control. If you do not see the Developer tab or the MCP key type, your Safeguard administrator has not turned on API key access to MCP for the workspace.

What to ask Claude once it is connected

Start with questions whose answers you can check against the Safeguard app. The default tools cover products, projects, vulnerabilities and findings, so these work on a new workspace:

List my products, then the projects in each one with their scan status.
Show critical vulnerabilities with a known exploit across my projects.
For the payments-api project, list findings and group them by package.
Get the full record for the first CVE in that list, including remediation.
Which of my projects have critical vulnerabilities with a fix version available?

Once your admin enables more tools, Safeguard's docs suggest prompts such as "Show me all SBOMs", "Generate a remediation plan for my npm project" and "Which packages violate my license policy?"

Two worked patterns:

Triage a release. Ask Claude to list critical and high vulnerabilities for one project, then ask for the affected package, installed version and fix version for each. Follow up with "which of these are in direct dependencies" so you know what your own manifest controls.

Ask what it cannot do. If Claude says a tool is not available, Safeguard returns a clear message naming the tool that is not enabled for your tenant. Copy that name and ask your Safeguard admin to enable it under Feature Flags.

Two habits keep answers honest. When a list's total is larger than what came back, ask Claude to page through the rest. And an empty result means nothing matched in your current scope, not that no vulnerability exists anywhere.

If you are deciding between assistants for coding work as well, our comparison of Claude and ChatGPT for coding covers how each runs and signs in.

Do the work properly with the Safeguard CLI

The MCP server brings Safeguard into Claude. The Safeguard CLI brings the work onto your own machine: ask questions in Claude, and write, fix and scan code in the terminal.

The Safeguard CLI scans code, container images and SBOMs on your own machine, sends only the results to the platform, and includes an assistant that can write code and fix what it finds.

Install

On macOS and Linux:

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

On Windows, in PowerShell:

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

It is a single static binary, with no Node, Python or JVM needed, for macOS, Linux and Windows on x86_64 or arm64 (Windows is x86_64 only). The download page lists version 1.4.0.

First session

Run safeguard with no arguments to open the interactive terminal. Type /login and your browser opens to sign you in. Then /scan . scans the folder you are in. After that, ask in plain language:

how many critical vulns do we have
fix the lodash CVE in this repo

Type @ to attach a file from the folder you started in. /resume picks up the last session in that folder.

Writing and fixing code

The assistant has tools for your local files and for your Safeguard data, and picks whichever the question needs. It can write files, edit code and run commands, with your approval. That makes it a general terminal assistant as well as a security one; the docs give "what's the latest file in ~/Downloads" as an example next to the vulnerability questions. Our post on the Safeguard CLI as an AI assistant on your machine goes deeper.

This demo attaches a computer running the Safeguard CLI and a folder, asks for a fix, reviews the diff and approves 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.

Using 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. /model shows or switches the AI model; see the CLI docs for connecting a model.

Approvals

Tools that change things ask before they act. /permissions sets the approval mode:

• Ask prompts before every action.
• Auto approves known actions and asks for anything new. This is the default.
• Yolo approves everything.

Start with Ask on a repository you care about.

Scan what changed, and run it in CI

safeguard scan --dir .
safeguard scan --repo https://github.com/your-org/your-repo
safeguard scan --image your-image:latest
safeguard scan --sbom ./sbom.cdx.json
safeguard validate ./sbom.cdx.json

Each scan builds an SBOM and matches it against vulnerability databases, writing JSON files such as vuln.json and license.json under ./output. Scan before and after any agent session, including one in Claude Code, and the difference is what that session did to your risk. Our guide to checking coding agent changes with the Safeguard CLI walks through it. safeguard validate checks CycloneDX, SPDX and VEX documents.

For CI, safeguard exec "summarise the critical findings in this repository" runs one non-interactive assistant turn. When the local runner runs platform scan jobs, your source code never leaves the machine; only SBOMs and findings are uploaded. Policies and gates are evaluated in the platform, which also offers reachability analysis and autonomous remediation with fix PRs, off by default.

One platform on every surface

Start in the terminal, keep asking in Claude, continue in the desktop app for Windows, macOS and Linux, and check findings on the iOS and Android apps. There are IDE extensions, a Chrome extension, and options with nothing to install: the web interface, the Cloud CLI, and a cloud sandbox tenant on a seeded demo repo, free for seven days with no card.

Download Safeguard, connect Claude, and run /scan . on the repository you work in today.

Common mistakes when connecting Claude

Editing claude_desktop_config.json. That file starts local servers. Safeguard's server is hosted, and Claude Desktop already uses the connectors in your Claude account.

Installing a package that claims to be the Safeguard MCP server. Safeguard publishes no npm package or local bridge for it. Treat any such package as untrusted.

Bringing your own OAuth client. Choose Register automatically. Safeguard accepts only client IDs it issued itself.

Clicking Authorize access without reading the page. Check that the consent page names Claude and claude.ai. If you did not start the connection, do not authorize it.

Expecting write actions on a new workspace. The default tool set is read only. Starting scans or opening fix pull requests needs an admin to allow write tools and enable those tools.

Using Allow always or Research with write tools on. Anthropic's article warns that Research can call connector tools without further approval. Turn write tools off first.

FAQ

Is Safeguard in Claude's connector directory?

Yes. Safeguard's download page says you can add it from Claude's connector directory on web, desktop or mobile, and Safeguard's Claude guide links to the directory entry. You need to be signed in to claude.ai for the link to open.

Which URL do I paste into Claude?

For a custom connector in claude.ai or Claude Desktop, Safeguard's Claude guide gives https://mcp.safeguard.sh/mcp/anthropic. The general endpoint, https://mcp.safeguard.sh/mcp, is the one Safeguard's setup page gives for Claude Code and most other clients.

Do I need an API key?

No. Claude signs in with OAuth and you click Authorize access on Safeguard's consent page. MCP keys exist only for clients that can send a fixed header and nothing else.

Which Claude plans can connect?

Anthropic's help center says custom connectors using remote MCP are available on Free, Pro, Max, Team and Enterprise, with Free users limited to one custom connector. On Team and Enterprise, an Owner adds the connector first and members then connect.

Can Claude change anything in my Safeguard account?

Only if your Safeguard admin allows it. A new workspace starts with six read-only tools, and with write tools off Claude can read but not change anything. Claude also asks for approval before it invokes tools.

How do I revoke Claude's access?

In Safeguard, open Settings › MCP connections and revoke the Claude connection. In Claude, remove or disconnect the connector under Customize › Connectors. Either one ends the access.

Can I switch to Safeguard and keep my Claude 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 keep using Claude itself with the Safeguard MCP server, as this guide shows.

Try it yourself

This walkthrough sets 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.

Quick check

Sources

• Get started with custom connectors using remote MCP, Claude Help Center
• Use connectors to extend Claude's capabilities, Claude Help Center
• Connect Claude Code to tools via MCP, Claude Code docs
• Safeguard MCP: Claude guide
• Safeguard MCP: connector setup
• Safeguard MCP overview
• Safeguard MCP: tool availability and gating
• Safeguard MCP setup
• Use Safeguard from Claude, interactive demo
• Safeguard download page
• Safeguard CLI docs
• Safeguard CLI interactive terminal docs

Last reviewed: 21 October 2026.