frontend-test
Fail
Audited by Gen Agent Trust Hub on Jun 22, 2026
Risk Level: HIGHCREDENTIALS_UNSAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [CREDENTIALS_UNSAFE]: The skill instructions explicitly direct the agent to retrieve sensitive authentication data, specifically an 'API_KEY', from the environment configuration. This involves accessing the '.env' file, which is a sensitive file path used for storing private secrets.
- [COMMAND_EXECUTION]: The skill employs powerful automation tools, including 'Bash' and browser tools capable of executing arbitrary JavaScript ('mcp__claude-in-chrome__javascript_tool'), to interact with and modify the application state.
- [PROMPT_INJECTION]: The skill presents an indirect prompt injection risk because it processes untrusted data from the browser without adequate sanitization or boundary markers, while maintaining access to sensitive system tools. * Ingestion points: Browser page text and tab context data retrieved via 'mcp__claude-in-chrome' tools. * Boundary markers: Absent; there are no instructions to the agent to treat data from the browser as untrusted or to ignore embedded instructions. * Capability inventory: Shell access ('Bash'), file system access ('Read'), and arbitrary JavaScript execution within the browser ('javascript_tool'). * Sanitization: Absent; the agent is not instructed to validate or filter content retrieved from the web environment before taking action.
Recommendations
- AI detected serious security threats
Audit Metadata