ad-creative-intelligence
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data by scraping ad copy and landing page content from Meta and Google. This creates a surface for indirect prompt injection, where malicious instructions hidden in advertisements could attempt to influence the agent's behavior during analysis.
- Ingestion points: Ad headlines, body text, and landing page descriptions fetched via the
meta-ad-scraperandgoogle-ad-scraperscripts referenced inSKILL.md. - Boundary markers: The skill uses structured formats like markdown tables and blockquotes to organize the data, but lacks explicit instructions to the agent to disregard potential instructions within the scraped text.
- Capability inventory: The agent is authorized to execute shell commands to run scraping scripts and write report files to the local file system.
- Sanitization: No specific sanitization or filtering logic is described for the retrieved ad content before it is interpreted by the agent.
- [COMMAND_EXECUTION]: The skill executes shell commands to run local Python scripts (
python3 skills/meta-ad-scraper/scripts/scrape_meta_ads.py). This behavior is consistent with the skill's purpose as a composite tool for ad intelligence.
Audit Metadata