Marshall Goodman

Building AI Security Against Threats

About

I started StrategicPromptArchitect (SPA) because the files and prompts we feed AI systems can carry hidden instructions that hijack how they behave. Prompt injection is now the number one risk for LLM applications (OWASP Top 10), and the attack surface keeps growing. I've built three tools to address it: Sentinel.AI, a browser extension that warns about risky prompts in real time; MalPromptSentinel, which extends detection to Claude Code skill files; and MPS-Agentic, a cloud platform that scans files across 26 attack categories using zone-weighted risk scoring. Think of it as antivirus for AI — scanning inputs before your AI processes them, in places traditional security doesn't look. All 3 are available on my site.