lachydotmcg/slopsec ? reverse-engineered prompt

Build me a Claude Code skill that helps me security audit vibe coded SaaS apps before launch. I want it to turn common app security mistakes into a simple repeatable review, so I can scope the app, walk through a checklist, confirm real findings, score them by severity, and then generate a clear report with fixes and a recheck step.

Make it practical for non experts, focused on defensive, authorized reviews only, and centered on the usual problem areas like secrets, auth, databases, injections, session handling, frontend trust, rate limits, AI specific risks, logging, and dependencies. Include a checklist I can follow top to bottom, a severity system that makes the most dangerous issues stand out, and a clean findings report template.

I’d also like it packaged so I can install it as a Claude Code skill or plugin and run it with a simple command. If you need to, look up current Claude Code skill docs online first.