brainstorming

Pass

Audited by Gen Agent Trust Hub on Jul 13, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes local bash scripts (start-server.sh, stop-server.sh) and starts a Node.js server (server.cjs) to provide an interactive visual companion in the user's browser. This is used to display design mockups and gather structured feedback.
  • [DATA_EXPOSURE]: The local server provides access to files within a designated session directory. Access is protected by a session-key token generated at startup and verified via query parameters or cookies, preventing unauthorized local access to the design data.
  • [COMMAND_EXECUTION]: The server can launch the user's default browser to open the companion URL using standard system utilities like open, xdg-open, or rundll32.exe. This is a functional requirement for the visual interface.
  • [SAFE]: The skill uses robust security practices for its local server, including constant-time authentication token comparison, Cross-Origin Resource Policy (CORP), Content-Security-Policy (CSP) headers, and path traversal checks using realpath resolution.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 13, 2026, 04:13 PM
Security Audit — agent-trust-hub — brainstorming