The Agent Skills Directory

[COMMAND_EXECUTION]: The skill's core functionality is built upon the execution of local shell commands (e.g., ls, find, du, mv, mkdir, shasum). It utilizes string interpolation for user-provided directory paths (e.g., ls -la [target_directory]), which presents a surface for command injection if the directory names contain shell metacharacters and are not properly escaped by the executing environment.
[INDIRECT_PROMPT_INJECTION]: The skill analyzes external, untrusted data by scanning the local file system to identify file types and patterns.
Ingestion points: Directory paths and file metadata (filenames, sizes, and file type headers) are read into the agent's context (SKILL.md).
Boundary markers: The skill includes instructions to ask for user confirmation before significant actions, but it lacks specific delimiting markers to prevent instructions embedded in filenames or file content from influencing the agent's behavior during the 'analysis' phase.
Capability inventory: The skill has the capability to move, rename, and potentially delete files via shell commands across the file system (SKILL.md).
Sanitization: The skill explicitly instructs the agent not to follow symlinks outside of the authorized tree and to handle name conflicts gracefully, which serves as a partial mitigation against directory traversal and accidental data loss.

file-organizer