design-review

Pass

Audited by Gen Agent Trust Hub on Jul 15, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill downloads the Bun runtime installer from 'bun.sh' if it is not detected during the setup phase.
  • [REMOTE_CODE_EXECUTION]: Executes the Bun installation script directly from the network using a piped-to-shell command (curl | bash).
  • [COMMAND_EXECUTION]: Invokes several external binaries part of the 'gstack' toolset (e.g., gstack-config, gstack-update-check, gstack-telemetry-log) found in the user's home directory to manage configuration and state.
  • [COMMAND_EXECUTION]: Performs automated version control operations including 'git commit', 'git add', and 'git revert' to atomically apply and verify design fixes in the source code.
  • [COMMAND_EXECUTION]: Executes arbitrary JavaScript within a browser context via the '$B js' command to audit the DOM, extract design tokens, and evaluate performance.
  • [DATA_EXFILTRATION]: Transmits anonymous usage telemetry (including skill name, duration, and outcome) to an external endpoint. The skill explicitly prompts the user to opt-in or opt-out of this data collection.
  • [PROMPT_INJECTION]: Ingests and processes untrusted content from external web pages and local files like 'DESIGN.md', creating a surface for indirect prompt injection. This is a functional requirement of its role as a visual auditor.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 15, 2026, 12:53 PM
Security Audit — agent-trust-hub — design-review