The Agent Skills Directory

[SAFE]: The skill performs standard mathematical and statistical operations using well-known Python libraries such as pandas, numpy, and scipy.
[PROMPT_INJECTION]: The skill processes user-provided data files (CSV, Excel, JSON), which serves as an ingestion point for untrusted data. While this represents a potential surface for indirect prompt injection, the script treats inputs as data for calculation and does not attempt to parse or execute instructions contained within those files. The risk is mitigated by the tool's focus on numerical analysis.
[COMMAND_EXECUTION]: The skill includes a Python script intended to be run by the user or agent. The script uses standard argument parsing and file I/O for data processing without attempting to escalate privileges or perform unauthorized operations.

corr-insight