claude-in-chrome

Pass

Audited by Gen Agent Trust Hub on Aug 26, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection attacks because it ingests data from external, untrusted web sources.
  • Ingestion points: The agent is instructed to use mcp__claude-in-chrome__read_page, mcp__claude-in-chrome__read_console_messages, and mcp__claude-in-chrome__read_network_requests to gather information from the browser.
  • Boundary markers: There are no instructions for the agent to use delimiters or ignore embedded instructions within the retrieved web content.
  • Capability inventory: The agent can execute scripts (mcp__claude-in-chrome__javascript_tool), fill forms (mcp__claude-in-chrome__form_input), and navigate to arbitrary URLs.
  • Sanitization: The skill does not provide any mechanisms for sanitizing or filtering external content before it enters the agent's context.
  • [COMMAND_EXECUTION]: The skill provides access to mcp__claude-in-chrome__javascript_tool, which enables the execution of arbitrary JavaScript within the browser context. This allows for dynamic code execution on any page the agent navigates to.
  • [DATA_EXFILTRATION]: The tools mcp__claude-in-chrome__read_console_messages and mcp__claude-in-chrome__read_network_requests allow the agent to capture potentially sensitive technical data, including API keys, tokens, or session identifiers that may appear in logs or network traffic.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 26, 2026, 12:42 PM
Security Audit — agent-trust-hub — claude-in-chrome