USE INSIDE CURSOR · CLAUDE · WINDSURF

Run scans from your AI tool.

Once it’s set up, you can just ask Cursor or Claude things like “check my app for legal problems” — they’ll run the scan, show you what’s wrong, and apply the fixes themselves.

STEP 1 · GET A KEY

Sign up, get a key.

Free. The key works for one app at a time and only lets the tool run scans — it can’t change anything else.

cursor — your-app.comscan · read

cck_live_•••••••••••••••••••••••

STEP 2 · PASTE THIS

Paste this once.

Into your Claude or Cursor settings. Specifically: claude_desktop_config.json (for Claude) or the MCP section of Cursor’s settings.

claude_desktop_config.json+ 8 lines

{
  "mcpServers": {
    "comply-code": {
      "command": "npx",
      "args": ["-y", "@comply-code/mcp"],
      "env": { "CC_API_KEY": "cck_live_…" }
    }
  }
}

STEP 3 · JUST ASK

Then just ask.

You don’t learn new commands. Just ask Claude or Cursor what you want, in plain English. It picks the right tool itself.

claude · session 4f2a

You

Check the checkout page for ADA + pixels firing pre-consent.

Claude

Running via comply-code…

→ comply-code.scan(url="your-app.com")
← 8 findings (1 crit · 3 high · 4 med) · 14s

STEP 4 · APPLY & VERIFY

It fixes, then checks itself.

Your AI tool gets five things it can do. The two it uses most: run a scan, and re-scan after applying a fix to confirm it worked.

what your AI can do5 actions

scan          → check a URL for legal problems
list_findings → see what we found last time
get_fix       → get the exact fix instructions
re_scan       → check if a fix actually worked
diff_audit    → compare today's scan to last week's

PrivacyThe connection runs from your computer. We see your URL and what the page looks like publicly — nothing else. Your private code stays on your machine.