AI · Cybersecurity · Field Work

Before You Deploy Your Vibe-Coded App

A pre-deployment security walkthrough for the AI builder era — 9 categories, 50+ checks, plus a runnable AI Skill.

4 artifacts in this collection

Deep Read· Read it. Think with it.

What this form is: A long-form essay or playbook. Read it once, keep it as reference. Best for context, depth, and decision-making.

Long-form security walkthrough: 9 categories, 50+ checks, AI-specific risks like prompt injection and agent over-permission, EU AI Act intersection.

Prompt Pack· Copy-paste, ready.

What this form is: Copy-paste prompts you can drop into Claude, ChatGPT, or your tool of choice. Tested, tuned, and ready.

The exact prompt to drop into your AI assistant so it does the security review for you on real code.

AI Skill· Install it into your AI.

What this form is: An installable SKILL.md any AI agent (Claude Code, Cursor, Copilot) can load to gain a new capability.

Cross-agent SKILL.md format — works in Claude Code, claude.ai, Codex, Gemini CLI, Cursor. Install runs the review on your project automatically.

Toolkit· Clone it. Build with it.

What this form is: An open repository or runnable project — code, scripts, or templates you can clone and adapt.

Public GitHub repo: skill source, adapters for AGENTS.md / GEMINI.md, install script, test suite, CI. Fork and extend.

Other collections