systematic-debugging

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFE
Full Analysis
  • [Indirect Prompt Injection] (SAFE): The skill identifies a surface for indirect prompt injection by processing external error logs and stack traces. 1. Ingestion points: Error messages and logs (SKILL.md). 2. Boundary markers: Absent. 3. Capability inventory: Shell command execution (npm, bun, security, codesign) and file writing (SKILL.md). 4. Sanitization: Absent. This surface is inherent to debugging tasks and lacks malicious intent.
  • [Command Execution] (SAFE): The skill provides shell command templates for environment diagnostics and test runners. These are standard developer tools and are used as illustrative templates for the agent to follow in a controlled manner.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 05:58 PM