aesthetic-dreaming

Pass

Audited by Gen Agent Trust Hub on Feb 26, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill implements a structured design workflow focusing on user interaction and local file management. It does not attempt to access sensitive system files or credentials.
  • [COMMAND_EXECUTION]: The skill uses the Bash tool to execute 'open' commands. This is used solely to allow users to view the generated HTML test previews in their default browser, which is a standard functionality for a UI design tool.
  • [DATA_EXFILTRATION]: No network-enabled tools or operations are defined. The skill operates entirely within the local environment, reading configuration from '.claude/preferences.json' and 'PLUGINS.md' and writing to '.claude/aesthetics/'.
  • [PROMPT_INJECTION]: The instructions are prescriptive but lack any malicious override patterns. They are designed to ensure the agent follows the creative workflow correctly.
  • [INDIRECT_PROMPT_INJECTION]: The skill presents a surface for indirect injection as it ingests untrusted user input to generate templates and mockups.
  • Ingestion points: Phase 1 (initial description) and Phase 3 (follow-up answers).
  • Boundary markers: Absent; user input is merged directly into a context object for generation.
  • Capability inventory: 'Write' (file creation) and 'Bash' (opening generated previews).
  • Sanitization: Absent; the skill extracts concepts verbatim. However, the risk is minimal as the output is restricted to documentation and visual mockups.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 26, 2026, 08:53 AM