qe-visual-accessibility

Pass

Audited by Gen Agent Trust Hub on Oct 6, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it is designed to visit and process content from external websites (e.g., using vibium go to navigate to target URLs). Maliciously crafted content on these sites could theoretically attempt to override agent instructions.
  • Ingestion points: URLs passed to browser automation tools in SKILL.md and evals/qe-visual-accessibility.yaml.
  • Boundary markers: Absent; the skill does not explicitly instruct the agent to ignore instructions embedded in the target site's HTML.
  • Capability inventory: The skill uses vibium for browser control, node to run visual diff scripts, and jq for data processing.
  • Sanitization: No content sanitization or filtering logic is specified for the ingested HTML or text data.
  • [COMMAND_EXECUTION]: The skill utilizes local CLI tools (aqe, vibium) and executes Node.js scripts (e.g., visual-diff.js) to perform its primary functions. These operations are restricted to the local environment and the project's skill directory, representing standard tool orchestration.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 6, 2026, 12:12 PM