The Agent Skills Directory

[SAFE]: The skill is entirely focused on research methodologies and data analysis. The provided Python snippets for reliability analysis, factor analysis, and network analysis are standard industry practices.
[DATA_EXPOSURE]: The documentation references a hardcoded local file path (/Users/zhangmingda/clawd/.venv/bin/activate) which discloses a local username. This is a minor configuration leak with no direct security impact on the skill's execution.
[INDIRECT_PROMPT_INJECTION]: The content_analysis function provides a surface for processing untrusted text data.
Ingestion points: texts parameter in content_analysis (SKILL.md)
Boundary markers: Absent
Capability inventory: Data manipulation via pandas and Counter; no shell execution or network operations are present
Sanitization: Limited to case normalization via .lower().

social-science-analysis