Skills
Navigate
skills/seifbenayed/claude-code-sdk/browse/Gen Agent Trust Hub

browse

Pass

Audited by Gen Agent Trust Hub on Apr 13, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill processes untrusted web content, creating an indirect prompt injection surface.
  • Ingestion points: The get_state, screenshot, and extract actions in SKILL.md bring external data into the agent context.
  • Boundary markers: No explicit delimiters or instructions are provided to isolate untrusted web content from agent instructions.
  • Capability inventory: The Browser tool provides powerful capabilities including element interaction, file uploads, cookie management, and arbitrary JavaScript execution via evaluate.
  • Sanitization: No sanitization of the external web content is described before it is processed by the agent.
  • [COMMAND_EXECUTION]: The skill facilitates the execution of browser-specific commands and arbitrary JavaScript code. The evaluate action allows for dynamic execution within the browser context.
  • [DATA_EXFILTRATION]: Provides the ability to retrieve sensitive information from the browser. The cookies_get action and JavaScript evaluate (accessing localStorage) can be used to access session cookies and authentication tokens.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 13, 2026, 04:17 AM
Security Audit — agent-trust-hub — browse