video-asset-manager

Warn

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: MEDIUMCREDENTIALS_UNSAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [CREDENTIALS_UNSAFE] (MEDIUM): The script scripts/search_pexels.py accepts the Pexels API key via the --api-key command-line argument. This is considered unsafe as command-line arguments are often visible to other users and processes on the system (e.g., via ps or /proc).
  • [EXTERNAL_DOWNLOADS] (MEDIUM): The skill performs automated downloads from URLs retrieved from the Pexels API. While Pexels is a legitimate service, the script scripts/search_pexels.py blindly trusts the link field from the API response without validation, which could be exploited if the API response is intercepted or manipulated.
  • [INDIRECT PROMPT INJECTION] (MEDIUM): The skill possesses a vulnerability surface where untrusted data from scene-plan.json is used to construct prompts for GenSpark AI.
  • Ingestion points: scene-plan.json (specifically clipKeywords) processed in SKILL.md and scripts/search_pexels.py.
  • Boundary markers: Absent. The keywords are joined directly into search queries or AI prompts.
  • Capability inventory: File system write access (assets/ directory), Network GET requests (requests.get), and instructions for web-based AI interaction.
  • Sanitization: None. The skill does not filter or escape keywords before using them in external service calls or prompt construction.
  • [COMMAND_EXECUTION] (LOW): The skill documentation provides shell commands for running Python scripts. While standard, users should be cautious when executing scripts that take external files as input.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 16, 2026, 09:04 PM