Consultora-AMDT/claude-github-security-audit — reverse-engineered prompt
Reverse engineered prompt
Build me a Claude Skill that helps me decide if a GitHub repo is safe enough to use before I add it to my project.
I want to be able to ask something simple like, audit repo plus a GitHub URL, and get a clear risk report. Start by asking whether I actually need the repo, and if my existing stack already covers it, say SKIP. If it does need an audit, check project health, code security, supply chain risk, code quality, license issues, MCP or plugin attack surface when relevant, and outside trust signals.
The output should feel practical, not academic. Give a score, explain the biggest risks, show any automatic deal breakers like no license or hardcoded secrets, and finish with ADOPT, EVALUATE, CAUTION, or AVOID. Also include a quick audit mode and a full report mode, and save a clean HTML report with an executive summary, findings, and final recommendation.
Please include English and Spanish versions, with clear install and usage instructions for Claude.
Want more depth? Deep Reverse