Polaris Audit MCP Server

Run website audits directly from Claude, Cursor, or any MCP-compatible AI assistant.

Tools

Tool	Description
`audit_url`	Submit a URL for a free scan — returns a scan token
`get_scan_result`	Poll for results using the token from `audit_url`
`get_public_result`	Fetch a publicly shared report by UUID

Usage with Claude Desktop

Add to ~/Library/Application Support/Claude/claude_desktop_config.json:

{
  "mcpServers": {
    "polaris-audit": {
      "command": "npx",
      "args": ["-y", "@polarisaudit/mcp-server"]
    }
  }
}

Usage with Cursor

Add to .cursor/mcp.json in your project (or ~/.cursor/mcp.json globally):

{
  "mcpServers": {
    "polaris-audit": {
      "command": "npx",
      "args": ["-y", "@polarisaudit/mcp-server"]
    }
  }
}

Environment Variables

Variable	Default	Description
`POLARIS_API_BASE`	`https://polarisaudit.com/api/scanner`	Override the API base URL (useful for self-hosting)
`POLARIS_API_KEY`	(none)	Bearer token for authenticated endpoints

Example conversation

You: Audit https://example.com for me
Claude: [calls audit_url] The scan has been submitted. Token: abc123. Let me check when it's ready...
[calls get_scan_result after ~45s]
Here are the results for example.com:
Overall grade: B (72/100)
Privacy: 65/100 — cookie banner not detected
Security: 80/100
Accessibility: 74/100
Performance: 68/100
3 must-fix issues and 7 quick wins identified.

Development

npm install
npm run build
node dist/index.js   # runs the stdio server

polaris-audit