The 32-page security guide for people actually shipping AI-built apps in 2026 — plus a runnable Claude skill that audits your own code.
45% of AI-generated code contains OWASP Top 10 vulnerabilities. Hardcoded secrets. Missing input validation. Exposed API keys. SQL via string concat.
✅ The Big Seven AI-coded mistakes — pattern + fix for each
✅ Secrets management + the 5-minute leak playbook
✅ Auth — sessions vs JWT vs OAuth, the 4 JWT pitfalls
✅ SQLi · XSS · CSRF — modern-framework-aware
✅ Dependencies + the 6 HTTP security headers
✅ A 7-step pre-deploy audit
✅ Build a /security-audit skill — Claude audits its own code
✅ Curated security agents on GitHub worth following
✅ /security-audit + /dep-watch SKILL.md (drop into .claude/skills/)
✅ Deploy-gate checklist + .gitignore + pre-commit config
✅ Vercel + Netlify security-header configs, copy-paste ready
✅ Audit prompts + curated resources
Email toddybuilds@gmail.com — no questions.
🛡️ AUD $27 · one-time · you own it forever
— Toddy · mechatronics engineer · daily Claude Code user at a startup · @buildwithtoddy