ai-content-pipeline

Pass

Audited by Gen Agent Trust Hub on Aug 12, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The skill defines content pipelines where the output of one AI model (such as a script generated by Claude) is used as direct input for subsequent models (such as text-to-speech or video generation tools). This architecture introduces a surface for indirect prompt injection. \n
  • Ingestion points: Data from research steps or text generation models are saved to files and then read as input for downstream processing tools in SKILL.md. \n
  • Boundary markers: The provided bash scripts interpolate model outputs directly into command-line arguments or JSON payloads without using delimiters or instructions to ignore embedded commands. \n
  • Capability inventory: The skill uses the belt CLI to invoke various inference endpoints and media merger tools. \n
  • Sanitization: No sanitization or validation of the data flowing between pipeline stages is implemented. \n- [EXTERNAL_DOWNLOADS]: The skill provides instructions for installing external tools and dependencies. \n
  • The skill documentation references the installation of the belt-sh/cli skill and points to installation resources on the inference-sh GitHub repository.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 12, 2026, 10:39 PM
Security Audit — agent-trust-hub — ai-content-pipeline