The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses ls to list active teams in ~/.claude/teams/ and rm -rf to delete specific team directories during cleanup. These operations are intended and necessary for the core functionality of the skill.
[PROMPT_INJECTION]: The skill ingests team and member names from the local file system and configuration files to perform actions like deletion and messaging. While these values are used in file paths and commands, the skill incorporates a verification step (checking for the existence of config.json) which serves as a basic validation of the target path.
Ingestion points: Team names are derived from directory listings in ~/.claude/teams/ or provided via arguments; member names are read from config.json (SKILL.md).
Boundary markers: None present.
Capability inventory: File system operations (ls, mkdir, rm -rf), messaging (SendMessage), and resource management (TeamDelete) (SKILL.md).
Sanitization: The skill verifies the existence of a configuration file before deletion, but does not implement explicit input sanitization for path traversal on the team name variable.

team-stop