The Agent Skills Directory

[SAFE]: The skill operates entirely within a localized configuration directory (~/.claude/plugins/config/claude-for-legal-zh/commercial-legal/). Its operations (creating subdirectories for matters, writing markdown notes, and archiving folders) are standard administrative tasks for workspace management and do not exceed the expected scope of the skill.
[DATA_EXPOSURE]: The skill handles sensitive legal metadata (client names, case facts). However, analysis shows no network capabilities (no curl, fetch, or socket operations), meaning the data remains strictly local to the user's environment. There is no evidence of hardcoded credentials or unauthorized access to system-level sensitive files.
[INDIRECT_PROMPT_INJECTION]: The skill ingests user input to populate matter.md and history.md files. While these files are later read by the agent, the risk is mitigated by the lack of high-privilege capabilities. The skill does not execute arbitrary code or shell commands based on the content of these files, and it utilizes slugification (lowercase with hyphens) for file paths to prevent directory traversal.
[COMMAND_EXECUTION]: No shell commands, subprocess calls, or dynamic execution patterns were detected. The file manipulations are performed through the agent's internal file system interface rather than through external shell execution.

matter-workspace