competitor-ad-teardown
Pass
Audited by Gen Agent Trust Hub on Jul 10, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill fetches content from external websites, including competitor domains and landing page URLs found in ads, using the
fetch_webpagetool and specialized scraping scripts. - [COMMAND_EXECUTION]: The skill executes local Python scripts (
scrape_meta_ads.pyandscrape_google_ads.py) via the shell to interact with ad transparency libraries. These scripts are referenced as part of the skill's required infrastructure. - [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and analyzes content from arbitrary external landing pages. An attacker-controlled website could include instructions designed to mislead the agent's strategic analysis.
- Ingestion points: Content fetched from
[landing_page_url]in Phase 2 (SKILL.md). - Boundary markers: None explicitly defined in the instructions for processing fetched text.
- Capability inventory: Subprocess execution (Phase 1) and local file writing (Phase 4).
- Sanitization: No specific sanitization or filtering of external landing page content is described before the LLM performs the analysis.
Audit Metadata