web-scraper

Warn

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION] (MEDIUM): The skill processes untrusted data from external websites via fetch_page.py and extract_text.py. This creates an indirect prompt injection surface where malicious instructions embedded in scraped HTML or text could influence the agent's subsequent reasoning or actions.
  • Ingestion points: URL content fetched by fetch_page.py, extract_links.py, and extract_text.py.
  • Boundary markers: Absent. There are no instructions to the agent to treat scraped content as data rather than instructions.
  • Capability inventory: Network access (requests), file writing (--output flag), and potential for the agent to use extracted text for decision making.
  • Sanitization: Absent. The skill extracts raw HTML or text without filtering for potential injection patterns.
  • [DATA_EXFILTRATION] (LOW): The skill performs network operations to arbitrary URLs. While this is the intended purpose of a web scraper, network access is a prerequisite for data exfiltration.
  • [EXTERNAL_DOWNLOADS] (LOW): The skill requires the installation of external Python packages (requests, beautifulsoup4). These are standard and trusted libraries, though no versions are pinned in the requirements.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 16, 2026, 01:32 PM