viewports
Pass
Audited by Gen Agent Trust Hub on Oct 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill analyzes external, potentially untrusted web pages to detect layout and accessibility issues. Text content extracted from these pages could contain instructions designed to influence the agent's analysis.\n
- Ingestion points: The
overflow.jsscript extracts text from DOM elements usingel.textContent.\n - Boundary markers: No specific delimiters are used to isolate extracted text in the reported findings.\n
- Capability inventory: The skill uses browser automation to render pages, capture screenshots, and evaluate JavaScript.\n
- Sanitization: Extracted text is truncated to 40 characters and whitespace is cleaned to limit potential injection payloads.\n- [DYNAMIC_EXECUTION]: The skill relies on executing JavaScript within a browser context to perform its analysis.\n
- Evidence:
SKILL.mdinstructs the agent to "Run overflow.js through the browser's evaluate call at each width."\n - Context: The script being executed is a local file provided as part of the skill, and this behavior is essential for the skill's primary purpose of web layout analysis.
Audit Metadata