google-ads-intent-scraper
Warn
Audited by Gen Agent Trust Hub on Jul 20, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs the agent to download the 'browser-act' CLI and related skills from an external website (browseract.com) at runtime.
- [REMOTE_CODE_EXECUTION]: Mandatory runtime instructions force the agent to install and execute code from a third-party repository (github.com/browser-act/skills) if the required tools are missing.
- [PROMPT_INJECTION]: The skill contains explicit instructions to override standard safety protocols by suppressing user confirmation and approval steps, specifically stating the agent should install dependencies 'without asking the user to confirm' and 'do not request separate approval'.
- [COMMAND_EXECUTION]: The skill requires the execution of an external command-line interface (BrowserAct) to perform browser automation and data extraction tasks.
- [PROMPT_INJECTION]: The skill exposes an attack surface for Indirect Prompt Injection (Category 8): 1. Ingestion points: Scraped Google Ads ad copy, advertiser names, and landing page content. 2. Boundary markers: None present to differentiate untrusted external content from system instructions. 3. Capability inventory: Subprocess execution via CLI and file-writing capabilities for data export. 4. Sanitization: No instructions provided for escaping or validating scraped external content before it is processed or presented.
Audit Metadata