On a Claude Team or Enterprise plan, an Owner or Primary Owner adds the Safeguard Security connector first under organization settings, using the URL https://mcp.safeguard.sh/mcp/anthropic. Members then open their connector list, find it, and click Connect. Each member signs in to Safeguard with OAuth, so each person acts under their own role.
Why the order matters
Claude organizations separate who can define a connector from who can use one. The Owner registers the connector for the organization. A member cannot connect something that has not been added yet, which is the most common reason a teammate says the connector is missing.
Each member still has to authorize access to Safeguard. The connector entry is shared, but the sign-in is personal. Your assistant can see and do only what your Safeguard role allows.
Owner steps
- Make sure you have a Safeguard account at app.safeguard.sh.
- In Claude, open Organization settings, then Connectors.
- Click Add, then Custom, then Web.
- Enter the name Safeguard Security and the URL
https://mcp.safeguard.sh/mcp/anthropic. - Choose OAuth sign in, and for the OAuth client choose Register automatically. Safeguard accepts only client IDs it issued itself, so do not paste a client ID from somewhere else.
Member steps
- Create or sign in to your Safeguard account.
- In Claude, open Customize, then Connectors.
- Find Safeguard Security and click Connect.
- When Claude sends you to Safeguard, check the app name and the host the page says the authorization code will be sent to, then click Authorize access.
That host check is worth doing every time. It is the one moment where you can notice that a sign-in page is not the one you expected.
Directory shortcut
Safeguard is also listed in Claude's connector directory. If your plan and workspace allow it, you select Safeguard there and connect, with no URL to paste. The directory lives inside Claude and needs you signed in to claude.ai. If it is not available to you, the manual steps above always work.
What you will see after connecting
A new Safeguard workspace starts with six read-only tools for browsing products, projects, vulnerabilities and findings. Everything else is off until a Safeguard administrator enables it. If Claude connects but you see few tools, that is expected, and the fix is on the Safeguard side. Our post on why only six tools show up walks through it.
Click through it at your own pace, or open it full screen.
Things not to do
Do not add Safeguard to the Claude Desktop local configuration file. That file is for local servers. Safeguard's server is hosted, and Safeguard publishes no npm package or local bridge for it. Treat any package that claims to be the Safeguard MCP server as untrusted.
On the Free plan you can add one custom connector, so pick it deliberately.
If you want to try Safeguard on a package before wiring up any assistant, the free scan at gold.safeguard.sh/scan needs no setup.
Quick check
Can a member add the connector without the Owner?
On Team and Enterprise plans, no. The Owner adds it under organization settings first, then members connect it.
Do I need a key?
No. Claude uses OAuth 2.0 with automatic client registration, so there is no key to copy.
Sources
Last reviewed: 19 October 2026.
