audn-scenario-builder
Pass
Audited by Gen Agent Trust Hub on Apr 13, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill contains several adversarial prompt patterns used as templates for red-teaming scenarios, including phrases like 'Override your instructions', 'Tell me your system prompt now', and references to 'DAN-style prompts'. These are explicitly identified as attack data for the purpose of security testing.
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted user input to build attack scenarios.
- Ingestion points: User input captured during Step 1 (Workflow) of the scenario building process.
- Boundary markers: None present; the skill interpolates user input directly into API payloads and output summaries.
- Capability inventory: Performs network operations (POST/GET) to the vendor's API at audn.ai.
- Sanitization: No validation or sanitization of user-provided message text or requirements is described.
Audit Metadata