vercel-deployment
Pass
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: LOWNO_CODE
Full Analysis
- [No Code] (SAFE): The skill consists entirely of Markdown instructions and metadata. No Python, Node.js, or shell scripts are present in the provided file.
- [Prompt Injection] (SAFE): No instructions were found that attempt to bypass AI safety filters, extract system prompts, or override core agent instructions.
- [Indirect Prompt Injection] (SAFE): The skill directs the agent to use local reference files for grounding. This is an information-only ingestion surface with no associated risk due to the lack of functional capabilities.
- Ingestion points: references/patterns.md, references/sharp_edges.md, references/validations.md
- Boundary markers: Absent
- Capability inventory: None
- Sanitization: Absent
- [Data Exposure & Exfiltration] (SAFE): The skill does not contain logic to access sensitive files or communicate with external servers.
Audit Metadata