Safeguard for Claude.
Connect Safeguard’s MCP server to Claude by Anthropic and ask about your software supply chain security in natural language. Query vulnerabilities, SBOMs, packages, licenses, and compliance data — and let Claude reason over your real security posture.
This is Safeguard’s MCP server endpoint purpose-built for Claude. Add it as a custom connector, then authorize with OAuth or an API key.
What you can do in Claude.
Once the Safeguard Security connector is enabled, Claude can call 35+ security tools and read MCP resources and prompts on your behalf.
Query vulnerabilities & findings
Ask about CVEs, severities, and open findings across your projects in plain language — no query syntax required.
Explore SBOMs, packages & licenses
Inspect components, package versions, transitive dependencies, and license obligations pulled straight from your SBOMs.
AI-powered remediation plans
Generate concrete fix plans for npm, pip, Maven, Go, and Cargo projects — upgrade paths, breaking-change notes, and all.
Compare & analyze SBOMs
Diff and analyze SBOMs in both CycloneDX and SPDX formats to see what changed between builds and releases.
SCM integrations
Reach repositories across GitHub, GitLab, Bitbucket, and Azure DevOps to scope security questions to real code.
Risk scoring & compliance reporting
Surface risk scores and compliance posture, then turn the answers into reports your stakeholders can read.
Policy-gate enforcement
Evaluate policy gates for deployment readiness and understand exactly why a build would pass or be blocked.
Multi-tenant for enterprise
Built multi-tenant from the ground up, so large organizations can keep teams and data cleanly isolated.
Enable it in a few steps.
- 01
Create your Safeguard account
Register an account at app.safeguard.sh — or sign in if you already have one.
- 02
In Claude, add a custom MCP connector
Open Claude and add a custom MCP connector / server.
- 03
Enter the connector URL
Enter the URL https://mcp.safeguard.sh/mcp/anthropic and name it “Safeguard Security”.
- 04
Authenticate
Sign in to Safeguard and click Approve when Claude prompts you.
- ✓
Done — ask away
Safeguard’s 70+ security tools are now available to Claude. Start asking questions.
Prefer a CLI-based MCP client? Drop this into your MCP client config using mcp-remote.
{
"mcpServers": {
"safeguard": {
"command": "npx",
"args": ["-y", "mcp-remote", "https://mcp.safeguard.sh/mcp/anthropic"],
"env": { "SAFEGUARD_API_KEY": "<your-api-key>" }
}
}
}Replace <your-api-key> with a key from your Safeguard settings.
Auth is simple: you just need a Safeguard account. Create one at app.safeguard.sh if you don’t have one — then just Approve access when Claude asks. Prefer keys? Generate one at your API keys page.
Example prompts.
Bring your supply chain into Claude.
Add the Safeguard Security connector, authorize with OAuth or an API key, and start asking Claude about your vulnerabilities, SBOMs, and compliance in seconds.
Self-healing security runs on Safeguard.
Your first fix PR is minutes away.
No sales call required, even your agent can complete the purchase over MCP.