ecommerce-product-description-generator-api

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests text generated by external AI models, creating a potential surface for indirect prompt injection if the generated content contains malicious instructions.
  • Ingestion points: The content field retrieved from the task query endpoint in scripts/aigc_textgen.py.
  • Boundary markers: The script uses newline encoding (⏎) and the SKILL.md instructions emphasize distinguishing between facts and recommendations.
  • Capability inventory: The skill performs network requests via urllib and writes data/media files to the local file system.
  • Sanitization: Implementation includes newline encoding and a recursive redaction process for secrets in response payloads.
  • [EXTERNAL_DOWNLOADS]: The skill includes a utility function to fetch media from external sources.
  • Evidence: The download_media function in scripts/aigc_textgen.py allows downloading content from arbitrary http/https URLs provided in the task payload.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 04:14 AM
Security Audit — agent-trust-hub — ecommerce-product-description-generator-api