design-sketch

Pass

Audited by Gen Agent Trust Hub on Jun 23, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill directs the agent to inspect existing project files to match the visual language and terminology of the product. This behavior creates a surface for indirect prompt injection, where an attacker could place malicious instructions inside a project file that the agent then follows while generating the design sketch.
  • Ingestion points: Reads project files in the current working directory (SKILL.md).
  • Boundary markers: No specific delimiters or "ignore instructions" warnings are defined for the reading of project files.
  • Capability inventory: Directory creation and file writing of .html files (SKILL.md).
  • Sanitization: No explicit sanitization or content validation of the project data is mentioned.
  • [COMMAND_EXECUTION]: The skill performs file system operations, specifically creating a directory named ./.design_sketch/ and writing various .html files into it based on user requests. These operations are scoped to the project's root directory.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 23, 2026, 11:22 PM
Security Audit — agent-trust-hub — design-sketch