Launching today
Neural Sentinel
Open-source AI scanner. Map vulnerabilities to ISO 42001.
2 followers
Open-source AI scanner. Map vulnerabilities to ISO 42001.
2 followers
Neural Sentinel is a free, open-source GitHub Action that detects AI-native vulnerabilities (prompt injection, unconstrained tool execution) in <15s. Unlike cloud scanners, it performs local-execution analysis: zero code or prompts are sent to external SaaS APIs. It automatically maps findings to ISO/IEC 42001 and the EU AI Act. Recently achieved a 100% Passing score on the OpenSSF Best Practices badge.
Hey Product Hunt! 👋 I’m Mohamed, the founder and lead architect of Neural Sentinel.
The Problem: Everyone is building AI agents and RAG apps, but auditing them for security is still done manually. Legacy SAST tools miss AI-specific kill chains, and many cloud-based AI scanners require uploading your proprietary prompts to third-party SaaS APIs, creating massive data sovereignty risks.
The Solution: We built a privacy-preserving, local-execution GitHub Action that scans your AI code for vulnerabilities in <15s, mapping them directly to ISO/IEC 42001 and the EU AI Act. Zero code or prompts are sent to external SaaS APIs.
For teams with strict data residency requirements, we also offer a 100% air-gapped, self-hosted Enterprise Docker Appliance that runs local LLM inference entirely within your own VPC. The core GitHub Action recently achieved a 100% Passing score on the OpenSSF Best Practices badge.
I’d love to hear your feedback, answer any technical questions about the architecture, or hear what AI security features you wish existed. Let’s make AI safer together! 🕸️🛡️