Prompt library · Security
Stop an attack at Cloudflare before it reaches my server
A prompt for Claude, ChatGPT or Cursor with SudoWhizzy connected to your server.
Best for
A site behind Cloudflare (connected on the Sites page) that is slow or down under a flood of requests.
What you get
What Cloudflare sees (requests per hour, countries, what is already blocked), then rules at the edge: blocks for the heavy addresses or networks, a rate limit on the costly paths, a challenge for the worst country, under attack mode only if needed, each with its undo.
The prompt
Using SudoWhizzy, my site <domain> is behind Cloudflare and is being flooded.
Call cf_zones, then cf_analytics for the zone and tell me in plain words what is happening: requests per hour against the days before, where it comes from, which paths, and what the firewall already stops. Follow the bot-attack playbook (get_playbook), Cloudflare part first. Add an allow rule for my own address (<my address>) before anything else. Then block the heavy addresses or networks, rate-limit the search and login paths, and challenge a country only if I agree; always leave verified search engines out. Switch under attack mode on only if that is not enough, and tell me to switch it off later. If the server is connected too, make sure it reads the real visitor address from Cloudflare. End with every rule you made and its change_id, so cf_undo can remove it when the attack is over.
How to use it
- Connect your server to SudoWhizzy with the one-line install, and add your SudoWhizzy MCP address to your AI client (Claude, ChatGPT, Cursor or Claude Code).
- Paste the prompt and fill in the parts in angle brackets.
- Your AI looks at the server first, explains its plan, and asks for your approval before anything destructive. How approvals work.
Start free
Free connects one server read-only. Starter lets your AI make changes, with approvals, snapshots and rollback.
Connect your server