ai-wrapper-product
Pass
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: LOW
Full Analysis
- PROMPT_INJECTION (SAFE): The skill contains no instructions designed to override agent behavior or bypass safety filters. All content is instructional for software development.
- DATA_EXFILTRATION (SAFE): No hardcoded credentials, sensitive file paths, or unauthorized network operations were detected. Code snippets use standard placeholders and demonstrate usage tracking without data theft.
- EXTERNAL_DOWNLOADS (SAFE): References standard packages like
@anthropic-ai/sdkin example code, but does not perform automated downloads or execution from untrusted sources. - COMMAND_EXECUTION (SAFE): No dangerous shell commands, subprocess spawning, or dynamic execution (eval/exec) patterns were identified.
- INDIRECT PROMPT INJECTION (SAFE): While the skill teaches how to handle user input in AI wrappers, it is an informational resource and does not expose the current agent to untrusted data. It explicitly recommends input validation and sanitization as best practices.
Audit Metadata