skills/tt-a1i/hermes-agent/dogfood/Gen Agent Trust Hub

dogfood

Pass

Audited by Gen Agent Trust Hub on Aug 28, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill systematically interacts with and processes data from external websites, which presents a risk of indirect prompt injection.
  • Ingestion points: Untrusted data enters the context through browser_navigate (visiting URLs), browser_snapshot (reading DOM trees), browser_vision (visual content analysis), and browser_console (reading JavaScript execution logs).
  • Boundary markers: The instructions lack explicit delimiters or warnings to treat ingested web content as data only, increasing the risk that embedded instructions (e.g., in HTML comments or log messages) could influence agent behavior.
  • Capability inventory: The skill uses a suite of browser interaction tools (browser_click, browser_type, browser_press) and writes screenshots and markdown reports to the local file system.
  • Sanitization: There is no evidence of content sanitization or filtering for the data retrieved from the target applications.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 28, 2026, 12:47 PM
Security Audit — agent-trust-hub — dogfood