scraper-builder

Pass

Audited by Gen Agent Trust Hub on Oct 12, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill's primary function is to ingest and process data from arbitrary external websites (HTML and JSON), which constitutes a known attack surface for indirect prompt injection. Malicious actors could embed hidden instructions in target websites that might influence the agent's behavior when it processes the extracted data. \n
  • Ingestion points: External data enters the context via the Web Unlocker and Browser APIs as described in SKILL.md and references/site-analysis-guide.md. \n
  • Boundary markers: No explicit delimiters or instructions are provided in the scraper templates to isolate untrusted web content or to instruct the agent to ignore any embedded directives. \n
  • Capability inventory: The generated scrapers have access to network libraries (requests, aiohttp), browser automation (playwright), and file writing capabilities (json.dump), which could be leveraged if an injection attack were successful. \n
  • Sanitization: The skill uses structural parsing (CSS selectors and BeautifulSoup) to target specific data, but it does not implement content-level sanitization or safety filtering of the retrieved strings.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 12, 2026, 11:20 PM