nemo-guardrails
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill contains strings typically associated with prompt injection, such as 'Ignore all previous instructions' and 'Pretend you are DAN'. However, these are explicitly used within code examples to demonstrate how to configure jailbreak detection flows. They are part of the defensive configuration logic and do not represent an attempt to override the agent's behavior.
- [EXTERNAL_DOWNLOADS]: The skill references the official GitHub repository for NeMo Guardrails (github.com/NVIDIA/NeMo-Guardrails) and provides instructions to install the 'nemoguardrails' package from a standard registry. These are well-known resources associated with NVIDIA and are considered safe.
- [DYNAMIC_EXECUTION]: The skill demonstrates the use of
RailsConfig.from_contentto define safety flows using the Colang DSL. This is the intended functionality of the framework for processing safety rules and does not involve the unsafe execution of arbitrary code from untrusted sources.
Audit Metadata