brand-voice-ai-training
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill involves ingesting external brand samples (such as emails or social media posts) to be used as examples in an AI prompt template. This creates a potential surface for indirect prompt injection if the ingested samples contain adversarial instructions.
- Ingestion points: SKILL.md (Step 1: Capture the Voice Inputs, Step 3: Build the Brand Context Block).
- Boundary markers: The template uses structured section headers (e.g., VOICE, TONE, EXAMPLES OF OUR VOICE) to delimit instructions from sample data.
- Capability inventory: None. The skill provides workflow and text generation instructions without shell access or network capabilities.
- Sanitization: No explicit content sanitization or validation is defined for the ingested samples.
Audit Metadata