competitive-ads-extractor
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest data from external sources, specifically competitor advertisements from platforms like Facebook and LinkedIn. This creates a surface for indirect prompt injection attacks if an attacker places malicious instructions within their ad copy.
- Ingestion points: Web content scraped from Facebook Ad Library, LinkedIn, and other ad repositories as specified in the skill's extraction steps.
- Boundary markers: The instructions do not define specific delimiters or "ignore" directives to prevent the agent from accidentally following instructions embedded in the analyzed ad copy.
- Capability inventory: The skill implies access to web scraping tools, file system write access (e.g., saving to
~/competitor-ads/), and reasoning capabilities for categorization and insight generation. - Sanitization: No explicit sanitization or filtering of the external advertisement text is described before it is analyzed by the language model.
Audit Metadata