design

Pass

Audited by Gen Agent Trust Hub on Mar 1, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: Indirect Prompt Injection Surface. The skill ingests untrusted data from local configuration files which could influence downstream agent behavior during design generation phases.
  • Ingestion points: Reads brand personality and style preferences from ai/style.config.md and ~/.claude/style.config.md (SKILL.md).
  • Boundary markers: Absent; instructions do not specify delimiters or 'ignore' warnings for the loaded configuration content before use.
  • Capability inventory: Executes multiple automated design tools (/state-design, /frontend-design, /ui-revamp) and performs local file writes to ai/design-progress.md and the ./proposals/ directory.
  • Sanitization: Absent; configuration values such as 'brand words' are directly incorporated into the creative generation logic without validation or filtering.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 1, 2026, 09:37 AM