The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use shell commands for managing local configuration files and directories. This includes potentially destructive operations like rm -rf and moving/copying files within sensitive user home directory paths such as ~/.cursor, ~/.claude, and ~/.ssh (indirectly via tools like Aider).
[INDIRECT_PROMPT_INJECTION]: The skill possesses a vulnerability surface where instructions embedded in external files can influence the agent.
Ingestion points: The skill reads content from external skill files (SKILL.md and .md rules) and tool configurations (.aider.conf.yml, copilot-instructions.md) using the cat command.
Boundary markers: There are no boundary markers or delimiters defined to distinguish between data being managed and instructions for the agent.
Capability inventory: The skill provides full file system management capabilities including creating (mkdir, cat >), moving (mv), copying (cp), and deleting (rm) files.
Sanitization: There is no evidence of sanitization or validation performed on the content read from these external files before the agent processes them.

manage-skills