The Agent Skills Directory

[SAFE]: The skill serves as a structural planner and does not generate or execute runnable code, minimizing the risk of direct command injection or malicious execution.
[EXTERNAL_DOWNLOADS]: References to standard libraries such as numpy, pandas, matplotlib, and scipy appear in the documentation and examples. These are well-known, trusted packages for scientific computing and do not pose a security risk in this context.
[DATA_EXFILTRATION]: No network communication tools or methods (e.g., curl, wget) are present. The skill only defines paths for local workspace data and results.
[CREDENTIALS_UNSAFE]: No hardcoded secrets, API keys, or access to sensitive configuration directories (like .ssh or .aws) were identified.

model-code-analyzer