The Agent Skills Directory

Structured Local File Interaction: The skill reads configuration from and writes matter-specific work products to dedicated subdirectories within the ~/.claude/plugins/config/ directory. This usage appears consistent with standard plugin data management and context-maintenance practices.
External Information Retrieval: The skill is designed to interact with external legal research platforms and web search tools to provide up-to-date jurisdictional rules. It emphasizes the importance of verifying these sources and explicitly requires source attribution (e.g., tagging citations with [Westlaw] or [web search — verify]).
Automated Risk Assessment Logic: The skill defines a checklist of 'high-risk flags' (such as whistleblower status or protected leave) to assist the user in identifying potential legal complications before proceeding with a termination.
User Role Compliance Gates: The skill includes a conditional logic gate that checks the user's role. For non-lawyers, it restricts 'Clear to proceed' recommendations and instead provides a structured brief for attorney review, which is a standard safety and compliance best practice.
Data Ingestion and Provenance: While the skill processes external data from users and research tools, it implements a provenance system to track the source of all legal information, reducing the risk of accidental reliance on unverified or potentially injected content.

termination-review