design-adapt
Pass
Audited by Gen Agent Trust Hub on Nov 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to scan external codebase files (CSS, JS, Dart, etc.) and automatically implement design fixes. This ingestion of untrusted data creates a surface for indirect prompt injection where malicious instructions hidden in code comments or metadata could attempt to influence the agent's behavior.
- Ingestion points: Reads
package.json,pubspec.yaml, and all CSS/style files during the audit phase. - Boundary markers: None explicitly defined to distinguish between user instructions and data from the codebase.
- Capability inventory: The agent is authorized to modify files and run project build commands.
- Sanitization: No explicit sanitization or validation of the scanned file content is mentioned.
- [COMMAND_EXECUTION]: In Phase 7 (Self-Healing Validation), the skill instructs the agent to 'Run the project's build command'. While a standard development practice, executing arbitrary build scripts from an untrusted codebase carries inherent risks if those scripts are compromised.
Audit Metadata