NYC

tech-design-doc

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFE
Full Analysis
  • Prompt Injection (SAFE): No instructions to override safety filters, role-play as unrestricted AI, or extract system prompts were found.
  • Data Exposure & Exfiltration (SAFE): While the skill reads local project files to gather context (grep/glob), it does not perform network exfiltration or access sensitive system paths (e.g., ~/.ssh).
  • Unverifiable Dependencies & Remote Code Execution (SAFE): The skill contains no package installation commands or remote script executions.
  • Indirect Prompt Injection (LOW): The skill ingests untrusted codebase data to generate documentation. Evidence Chain: 1. Ingestion points: Local codebase files identified via grep/glob. 2. Boundary markers: Absent. 3. Capability inventory: File-write (outputting the generated design document). 4. Sanitization: Absent. As this behavior is central to the skill's primary purpose and involves low-risk output (markdown), the severity is classified as LOW.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 05:12 PM