marketing-ideas

Pass

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: LOW
Full Analysis
  • Indirect Prompt Injection (LOW): The skill directs the agent to read a local configuration file (.claude/product-marketing-context.md). This constitutes a data ingestion surface; however, because the skill has no capabilities for external communication (network) or side-effect actions (file writing/command execution), the risk of exploitation is negligible.
  • Data Exposure (LOW): The skill accesses a project-specific marketing context file. This is intended behavior for contextual assistance and does not target sensitive system files or credentials.
Audit Metadata
Risk Level
LOW
Analyzed
Feb 16, 2026, 09:11 AM