The Agent Skills Directory

File Persistence and Context Reading: The skill interacts with local files to retrieve matter-specific context and store expansion progress. Specifically, it reads from ~/.claude/plugins/config/claude-for-legal/employment-legal/CLAUDE.md and writes to ~/.claude/plugins/config/claude-for-legal/employment-legal/expansion-[slug].yaml. This persistent state is part of the tool's core functionality for tracking long-term projects.
Indirect Prompt Injection Surface: The skill processes information from external configuration files and user-provided inputs, which could potentially be manipulated to influence agent behavior. (1) Ingestion points: Context is gathered from CLAUDE.md, matter-specific workspace files (matter.md), and direct user intake responses in Step 1. (2) Boundary markers: The instructions do not specify explicit delimiters (like XML tags or clear boundaries) to separate external data from system instructions, which is a potential area for review. (3) Capability inventory: The skill possesses capabilities for reading local filesystem data and writing YAML-formatted tracker files. (4) Sanitization: There is no explicit mention of sanitizing or validating the input data before it is incorporated into the output headers or the persistent YAML files.

international-expansion