webconsulting-create-documentation
Pass
Audited by Gen Agent Trust Hub on Feb 18, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- PROMPT_INJECTION (LOW): The narration prompt template in 'narration-examples.md' interpolates a '[PRODUCT NAME]' variable, which creates a surface for indirect prompt injection if the source data is untrusted. * Ingestion points: The '[PRODUCT NAME]' placeholder within the prompt template used for Jony Ive-style narration. * Boundary markers: There are no explicit delimiters or safety instructions provided to the LLM to separate the dynamic product name from the rest of the narration rules. * Capability inventory: The documentation refers to a generation script that writes to 'remotion/narration-durations.ts' and uses external media APIs (ElevenLabs and Suno AI), which could be manipulated if the prompt is successfully injected. * Sanitization: No input validation or sanitization logic for the dynamic content is specified in the provided files.
Audit Metadata