CYBERSECURITYTRACKER
TRACKING6,506 stories in this site build1,309 vulnerability news stories in this site build
Permanent story citation

SingGuard-NSFA: Open-source guardrails for agentic AI

This page keeps the story as Cybersecurity Tracker first published it. If the tracker later corrects it, the correction appears below the original and never replaces it.

Back to newsStory 2546

As cited

Copy frozen at (site build).

ai security

SingGuard-NSFA: Open-source guardrails for agentic AI

SingGuard-NSFA is an open-source guardrail framework designed to detect operational threats in agent-based AI workflows. The framework includes four models ranging from 0.8B to 9B parameters and organizes risks using a taxonomy of 185 threat variants mapped to the CIA triad (confidentiality, integrity, availability) and validated against OWASP guidelines.

Why it matters: Security teams deploying agentic AI systems need practical detection mechanisms for prompt injection and other query-side threats; this framework provides an extensible starting point to identify risks before they escalate in production agent workflows.

Source published
First seen by Cybersecurity Tracker

Source attribution

Correction

Correction recorded as of .

ai security

SingGuard-NSFA: Open-source guardrails for agentic AI

No summary had been written when this copy was frozen.

First seen by Cybersecurity Tracker

Source attribution

Correction

Correction recorded as of .

ai security

SingGuard-NSFA: Open-source guardrails for agentic AI

SingGuard-NSFA is an open-source guardrail framework designed to mitigate operational threats in agent workflows, offering four models ranging from 0.8 billion to 9 billion parameters built on Qwen3.5 base models. The framework organizes security risks along confidentiality, integrity, and availability dimensions, defining 185 risk variants mapped to top-level domains and mid-level categories validated against OWASP guidelines.

Why it matters: Development teams deploying agentic artificial intelligence (AI) systems need guardrails to prevent prompt injection and other query-side threats that could compromise confidentiality, integrity, or availability of agent operations.

Source published
First seen by Cybersecurity Tracker

Source attribution

Glossary