Summary
The GitHub repository Nox-HQ/nox describes itself as an open-source security scanner with AI application security features, including checks for prompt injection, embedding leakage, agent over-privilege and MCP hardening. According to the project's description, it also produces polyglot AIBOM, SARIF and SBOM outputs and includes a Cosign-signed plugin marketplace. It is described as offline-first and agent-native via MCP. The repository shows 11 stars and a recent update signal, and the source does not provide further details.
AI-assisted summary based on the listed source.
What happened
Open-source security scanner with first-class AI app security (prompt injection, embedding leakage, agent over-privilege, MCP hardening). Polyglot AIBOM, SARIF, SBOM. Cosign-signed plugin marketplace. Offline-first, agent-native via MCP. Stars: 11. Updated repository signal.
Why it matters
Developers and security teams building LLM apps and AI agents could use a scanner like this to look for the specific risks listed, if it works as described. Teams that already work with SARIF or SBOM outputs may find the format fit familiar. Because the description is self-reported and the project has only 11 stars, readers should watch for independent testing, release history and wider adoption before relying on it.
Signal Intelligence
Signal Strength 95%
Technical label SOURCE-BACKED
Public Interest 30
Category SECURITY
Reader Depth TECHNICAL
Signal Strength reflects source quality, relevance, freshness and evidence. Public Interest helps organize discovery; it is not proof of truth.
Public Interest components
Recognizable Entity Score 0
Practical Impact Score 24
Novelty Interest Score 72
Consequence Score 46
Curiosity Score 16
Shareability Score 25