decision-toolkit

Pass

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: LOWPROMPT_INJECTION
Full Analysis
  • Indirect Prompt Injection (LOW): The skill templates interpolate user-provided data into markdown and voice scripts. While this creates a surface for indirect instructions, the skill has no capability to execute commands, access the network, or make automated decisions.
  • Ingestion points: Template placeholders in decision-export-template.md, decision-framework.md, and decision-voice-summary.md.
  • Boundary markers: Absent; data is directly interpolated into templates.
  • Capability inventory: Markdown generation and browser-based file download. No network or filesystem access.
  • Sanitization: No sanitization of user-provided strings before interpolation.
Audit Metadata
Risk Level
LOW
Analyzed
Feb 16, 2026, 12:36 AM