The Agent Skills Directory

[SAFE]: The skill provides educational content and functional code snippets for academic statistical research. It correctly references well-established Python libraries and follows professional reporting standards. There is no evidence of command injection, data exfiltration, or malicious intent.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied datasets, which establishes an attack surface for indirect prompt injection, although the risk is minimal given the lack of dangerous capabilities.
Ingestion points: Input parameters for dataframes (data, df) identified in the analysis code blocks in SKILL.md.
Boundary markers: No explicit delimiters or boundary instructions are used to isolate user data from the analysis logic.
Capability inventory: The skill utilizes scipy, statsmodels, pingouin, pymc, and matplotlib in SKILL.md for statistical computation and visualization. It does not possess capabilities for network operations, shell command execution, or unauthorized file system access.
Sanitization: No specific sanitization or validation logic is defined for the input data streams.

statistical-analysis