The Agent Skills Directory

[SAFE]: The skill is designed for academic auditing of statistical claims. It provides a structured framework for evaluating the validity of causal language, generalizations, and interpretations of uncertainty within a social science context.
[COMMAND_EXECUTION]: The skill utilizes a local Python script (scripts/claim_audit.py) via the uv run command to perform automated linting of inferential claims. The script is restricted to the Python standard library and functions by comparing user-provided claim sentences against predefined regular expression patterns that identify common statistical misinterpretations.
[INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided text (draft academic claims) for auditing purposes. Analysis of the ingestion surface shows it is safe, as the associated tooling is limited to providing informational feedback to the agent and user. It does not possess capabilities for network operations, sensitive file system access, or execution of unsanitized external content.

alterlab-ssci-inference-gate