Skip to main content

prompt-guard

Detects prompt injections and jailbreak attempts in LLM applications, ensuring secure and safe user interactions.

Install this skill

or
69/100

Security score

The prompt-guard skill was audited on Jun 11, 2026 and we found 3 security issues across 3 threat categories, including 2 high-severity. Review the findings below before installing.

Categories Tested

Security Issues

high line 31

Eval function call - arbitrary code execution

SourceSKILL.md
31model.eval()
high line 42

Prompt injection: ignore instructions

SourceSKILL.md
42score = get_jailbreak_score("Ignore previous instructions")
low line 308

External URL reference

SourceSKILL.md
308- **Model**: https://huggingface.co/meta-llama/Prompt-Guard-86M
Scanned on Jun 11, 2026
View Security Dashboard
Installation guide →