Model Context Protocol

Safeguard MCP Server

Connect your AI assistant directly to your software supply chain security data — query vulnerabilities, SBOMs, findings, and compliance, and run scans in plain language.

MCP URL https://www.mcp.safeguard.sh/mcp

There is nothing to install: this server is hosted. Clients that support OAuth need no key. For a platform that can only send a fixed header, create an MCP key in Safeguard under Settings › Developer (key type MCP key) and send it as Authorization: Bearer sg_sk_.... MCP keys are offered only where API-key access to MCP is on for your workspace; ask your Safeguard administrator. The key acts as you, with your MCP access, and stops working within a minute of being revoked.

Connect to Claude (claude.ai and Claude Desktop)

Add Safeguard to Claude as a custom connector. It signs in with your Safeguard account over OAuth 2.0 (authorization code with PKCE), so there is no key to copy and nothing to install.

Connector URL https://www.mcp.safeguard.sh/mcp/anthropic
1
In Claude, open Customize › Connectors, click + Add, then Add custom connector. On a Team or Enterprise plan, an Owner adds it under Organization settings › Connectors, and members then find it in Customize › Connectors and click Connect.
2
Name it Safeguard, paste the connector URL, and click Continue.
3
Under Authentication choose Sign in now, and under OAuth client choose Register automatically. Click Add.
4
Sign in to Safeguard, check the app name and the host the authorization code will be sent to, then click Authorize access.

Claude Desktop uses the connectors you add in Claude, so the same connector works there. claude_desktop_config.json only starts local servers; Safeguard is hosted, so nothing goes in that file.

Connect to ChatGPT (OpenAI)

ChatGPT adds Safeguard as a custom MCP server and registers itself with Safeguard's OAuth server (POST /register, RFC 7591), so there is no client ID or key to copy. Your ChatGPT account and workspace policies decide whether you can add a custom MCP server.

MCP URL https://www.mcp.safeguard.sh/mcp
1
In ChatGPT, open Plugins, select the + button, then Add custom MCP server.
2
Enter a name and description. Under Connection, paste the MCP server URL.
3
Set authentication to OAuth (ChatGPT registers itself with Safeguard), review the risk warning, select I understand and want to continue, then select Create as a plugin.
4
Sign in to Safeguard, check the app name and the host the authorization code will be sent to, then click Authorize access.

Connect to Google Gemini

Option A — Gemini CLI (recommended for developers). Add to ~/.gemini/settings.json:

{
  "mcpServers": {
    "safeguard": {
      "httpUrl": "https://www.mcp.safeguard.sh/mcp",
      "timeout": 30000,
      "headers": {
        "Authorization": "Bearer YOUR_MCP_KEY"
      }
    }
  }
}

Option B — Google AI Studio (web). Uses the same OAuth flow as ChatGPT:

1
Open aistudio.google.com → Tools & Extensions → Add MCP Server
2
Paste: https://www.mcp.safeguard.sh/mcp and follow the OAuth login prompt.

Download ready-made Gemini CLI config →

Connect to Cursor IDE

Cursor connects to remote MCP servers and signs in with OAuth, so the entry is the URL alone. Add it to ~/.cursor/mcp.json (global) or .cursor/mcp.json (per project):

{
  "mcpServers": {
    "safeguard": {
      "url": "https://www.mcp.safeguard.sh/mcp"
    }
  }
}

The first time Cursor connects, it opens Safeguard in your browser. Sign in to Safeguard, check the app name and the host the authorization code will be sent to, then click Authorize access.

Download ready-made Cursor config →

Connect to Google Vertex AI

Use Safeguard with Google Vertex AI agents and the Vertex AI Agent Builder on cloud.google.com.

{
  "mcpServers": {
    "safeguard": {
      "url": "https://www.mcp.safeguard.sh/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_MCP_KEY"
      }
    }
  }
}

For Vertex AI Agent Builder: add this as an external tool in your agent configuration. For Vertex AI extensions, use the OpenAPI spec at https://www.mcp.safeguard.sh/openapi.json.

Download Vertex AI config →  |  Create an MCP key under Settings › Developer

Connect to Google NotebookLM

Google NotebookLM supports MCP tool integrations to augment research with live security data.

1
Go to notebooklm.google.com → Settings → Integrations → Add MCP server
2
Paste: https://www.mcp.safeguard.sh/mcp and authorize with Safeguard.

Download config →

Connect to Microsoft Copilot (Bing)

Microsoft Copilot at bing.com/copilot and copilot.microsoft.com supports MCP servers.

1
Go to copilot.microsoft.com → Settings → Plugins & Integrations → Add MCP server
2
Paste: https://www.mcp.safeguard.sh/mcp and sign in with Safeguard.
3
Ask Copilot to "show my security vulnerabilities" or "check my SBOM compliance".

Download config →

Connect to Azure AI Foundry / Azure OpenAI

Use Safeguard with Azure AI agents and Azure OpenAI via ai.azure.com (Azure AI Foundry).

{
  "mcpServers": {
    "safeguard": {
      "url": "https://www.mcp.safeguard.sh/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_MCP_KEY"
      }
    }
  }
}

For Azure AI Foundry agents: add as an MCP tool server in your agent's tool configuration. For Azure OpenAI Assistants: use the OpenAPI spec at https://www.mcp.safeguard.sh/openapi.json.

Download Azure config →  |  Create an MCP key under Settings › Developer

Connect to Meta Llama

Use Safeguard with Meta's Llama models via llama.meta.com and Llama API integrations.

1
Go to llama.meta.com → Developer tools → MCP Servers → Add server
2
Paste: https://www.mcp.safeguard.sh/mcp and authorize with Safeguard.
3
Alternatively, use the MCP URL directly in any Llama-compatible agent framework.

Download config →

Connect to Perplexity AI

Perplexity supports MCP servers natively via OAuth 2.0. Connect directly from your Perplexity settings.

1
Go to perplexity.ai → Settings → MCP Servers → Add server
2
Paste the MCP URL: https://www.mcp.safeguard.sh/mcp
3
Perplexity will redirect you to Safeguard to sign in. Check the app name and the host the authorization code will be sent to, then click Authorize access.
4
Ask Perplexity to "show my critical vulnerabilities" or "list my security projects".

Download config →  |  Create an MCP key under Settings › Developer

Connect to Grok (xAI)

Grok supports MCP servers on grok.com. Connect via the integrations panel.

1
Go to grok.com → click your profile → Settings → Connected tools → Add MCP server
2
Paste: https://www.mcp.safeguard.sh/mcp
3
Sign in to Safeguard when prompted, check the app name and the host the authorization code will be sent to, then click Authorize access.
4
Ask Grok to analyze your software supply chain security.

Download config →

Connect to VS Code (GitHub Copilot)

GitHub Copilot in VS Code uses the MCP servers in VS Code's MCP configuration. VS Code connects to remote servers over HTTP and runs Safeguard's OAuth sign-in itself. Add this to .vscode/mcp.json in your workspace, or to your user configuration with the MCP: Open User Configuration command:

{
  "servers": {
    "safeguard": {
      "type": "http",
      "url": "https://www.mcp.safeguard.sh/mcp"
    }
  }
}

When the server starts, VS Code opens Safeguard in your browser. Sign in to Safeguard, check the app name and the host the authorization code will be sent to, then click Authorize access.

Download ready-made VS Code config →

Another assistant?

Safeguard's MCP endpoint is the same for every client: https://www.mcp.safeguard.sh/mcp. A client that supports OAuth signs in with your Safeguard account; a platform that can only send a fixed header sends an MCP key as Authorization: Bearer sg_sk_…. Nothing needs installing.

See setup steps for all supported assistants →

Available Endpoints

POST/mcpMCP JSON-RPC endpoint (all clients)
GET/mcpCapability discovery
POST/registerOAuth Dynamic Client Registration (RFC 7591) — ChatGPT, Gemini
GET/.well-known/oauth-authorization-serverOAuth 2.0 server metadata (RFC 8414)
GET/.well-known/oauth-protected-resourceProtected resource metadata (RFC 9728)
GET/authorizeOAuth authorization endpoint
POST/tokenOAuth token exchange
GET/userinfoOAuth user info
GET/claude-desktop-configClaude connector URL and steps (JSON)
GET/gemini-configGemini CLI config JSON
GET/vertexai-configGoogle Vertex AI config JSON
GET/notebooklm-configGoogle NotebookLM config JSON
GET/cursor-mcp-configCursor MCP config JSON
GET/bing-configMicrosoft Copilot (Bing) config JSON
GET/azure-configAzure AI Foundry config JSON
GET/llama-configMeta Llama config JSON
GET/perplexity-configPerplexity config JSON
GET/grok-configGrok (xAI) config JSON
GET/copilot-configVS Code (GitHub Copilot) config JSON
GET/setup/allSetup steps and config downloads for every supported assistant
GET/openapi.jsonOpenAPI 3.0 schema
GET/healthHealth check