develop-securely
Run threat modeling, an OWASP-based secure coding checklist, and severity-classified security review across a software or AI-agent change. Use when asked to threat model a design, secure code against vulnerabilities, review a diff or PR for security issues, harden an AI agent/LLM/MCP/tool-use surface, triage a vulnerability report, or when a non-technical vibe-coder asks something like "is my app safe," "can I launch this," "check this before I publish it," or "這個安全嗎" / "我要上線了幫我看一下" before showing an AI-built app to real users. Produces a scoped threat model, applies the relevant checklist, and delivers findings with exploit reasoning and remediation status — not a rubber-stamp pass — rendered in the language and depth the requester can act on.
