skills/unbrowse-ai/sites/gov-uk/Gen Agent Trust Hub

gov-uk

Pass

Audited by Gen Agent Trust Hub on Oct 7, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill retrieves content from gov.uk, which represents an external data ingestion surface. While gov.uk is a primary government domain, processing its content without strict isolation poses a theoretical risk of indirect instructions. \n- Ingestion points: Content is fetched through gov_uk__get_search_all, unbrowse.scrape, and unbrowse.map (SKILL.md). \n- Boundary markers: The instructions do not provide explicit delimiters or warnings to separate external web content from the agent's instructions. \n- Capability inventory: The agent has capabilities to perform tasks and browser actions on gov.uk via unbrowse.run and unbrowse.browse.act (SKILL.md). \n- Sanitization: No specific methods for sanitizing or filtering retrieved web data are described.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 7, 2026, 04:13 AM