The Agent Skills Directory

[SAFE]: The skill provides health checks by reading platform-specific files such as session monitor states and configuration hooks in ~/.claude/. This behavior is aligned with the primary purpose of session maintenance.
[SAFE]: Tool use is restricted to reading and listing files (Read, Glob, and Bash limited to ls and cat), which prevents any unauthorized modification of the underlying system.
[SAFE]: The skill identifies an indirect prompt injection surface by ingesting data from log files and focus documents. However, the risk is negligible as the available tools are restricted to basic file reading, and no network exfiltration capabilities are present.

context-status