The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes several helper scripts (create_subagent.py, delete_subagent.py, list_subagents.py, move_subagent.py) to perform filesystem operations. These scripts are used to manage files in standardized configuration paths such as ~/.claude/agents/ and .claude/agents/.
[DATA_EXPOSURE]: The skill interacts with sensitive configuration directories that store agent instructions and system prompts. This access is restricted to the intended purpose of managing subagent definitions.
[INDIRECT_PROMPT_INJECTION]: The skill provides a mechanism to create persistent subagent definitions which include system prompts, creating a surface for potential instruction injection if the input is untrusted.
Ingestion points: scripts/create_subagent.py accepts a --prompt argument used to populate the system prompt of a new subagent.
Boundary markers: No boundary markers or 'ignore embedded instructions' warnings are present in the generated files.
Capability inventory: The skill possesses capabilities for file creation (create_subagent.py), file deletion (delete_subagent.py), and file movement (move_subagent.py).
Sanitization: The create_subagent.py script validates the subagent name for alphanumeric characters but does not sanitize or escape the content of the system prompt.

subagents-management