The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of powershell.exe -Command and WSL2 shell commands to perform disk operations like Remove-Item for file deletion, wsl --unregister for system migration, and Optimize-VHD for disk compaction.- [DATA_EXFILTRATION]: Facilitates the movement of sensitive local files, such as SSH keys, GPG keys, and .env configuration files, to external cloud storage providers using the rclone utility.- [EXTERNAL_DOWNLOADS]: Recommends and executes the installation of third-party tools like npkill, dust, gdu, and rclone using apt and npx package managers during runtime operations.- [COMMAND_EXECUTION]: Shell scripts provided for maintenance tasks (e.g., the find loop for Git garbage collection) process directory names without proper quoting or sanitization, creating a potential command injection surface if file paths contain shell metacharacters.- [DATA_EXFILTRATION]: Accesses and processes sensitive directories including ~/.ssh/ and .env.example as part of its backup prioritization tiers.- [COMMAND_EXECUTION]: Category 8 Evidence Chain: 1. Ingestion points: Local file system directory names and paths (SKILL.md). 2. Boundary markers: Absent; the skill does not instruct the agent to ignore instructions embedded in data. 3. Capability inventory: Remove-Item, git gc, rclone, powershell.exe execution (SKILL.md). 4. Sanitization: Absent; shell variables in maintenance loops are not escaped or quoted to prevent injection.

storage-management