url-dump

Pass

Audited by Gen Agent Trust Hub on Mar 14, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: Indirect Prompt Injection risk via external content ingestion. The skill processes untrusted data from user-provided URLs to generate summaries and categorization, creating a surface for embedded instructions to influence the agent.\n
  • Ingestion points: Web content retrieved through the web-fetch integration (SKILL.md).\n
  • Boundary markers: Absent. The instructions do not specify delimiters or warnings to ignore commands within the fetched data.\n
  • Capability inventory: Filesystem write access to various vault directories including 00-inbox/, 04-projects/, and 05-knowledge/ (SKILL.md).\n
  • Sanitization: Absent. Content is analyzed directly for executive summaries and insights without validation or escaping.\n- [EXTERNAL_DOWNLOADS]: The skill performs automated retrieval of external data using the web-fetch integration. While necessary for its core purpose of URL archiving, this involves downloading untrusted content into the agent's environment.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 14, 2026, 11:30 AM