ai-podcast

Pass

Audited by Gen Agent Trust Hub on Aug 31, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill documents a legitimate generative AI pipeline for creating multi-speaker videos. The use of 'infsh app run' commands points to internally managed inference applications rather than arbitrary external scripts.
  • [INDIRECT_PROMPT_INJECTION]: The skill has a surface area for indirect prompt injection as it ingests user-supplied text and images to generate media.
  • Ingestion points: Subject images for Phota training and script text for TTS generation in SKILL.md.
  • Boundary markers: The instructions do not define specific delimiters for user-provided data within the shell command payloads.
  • Capability inventory: The skill uses Bash for controlling the inference pipeline and Write for saving configuration files.
  • Sanitization: The skill relies on the safety filters of the underlying generative models (e.g., Google Gemini, Inworld) to handle untrusted input.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 31, 2026, 11:00 PM
Security Audit — agent-trust-hub — ai-podcast