ad-creative-intelligence

Pass

Audited by Gen Agent Trust Hub on Jul 10, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: Indirect Prompt Injection Risk.
  • Ingestion points: Ad headlines, body text, descriptions, and landing page URLs scraped from Meta and Google Ad Libraries (Phase 1, Phase 2).
  • Boundary markers: None present. The instructions do not specify delimiters or warn the agent to ignore instructions embedded within the scraped ad data.
  • Capability inventory: The skill executes subprocesses via shell commands (python3 skills/...) and performs complex clustering and analysis logic on the ingested data (Phase 3).
  • Sanitization: No explicit sanitization, validation, or instructions to treat scraped content as untrusted data are provided.
  • [COMMAND_EXECUTION]: The skill uses shell commands to execute local Python scripts (scrape_meta_ads.py and scrape_google_ads.py) found in relative paths within the skill structure to perform its primary scraping functions.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 10, 2026, 03:06 PM
Security Audit — agent-trust-hub — ad-creative-intelligence