The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute the claude agents CLI command to programmatically identify available personas from the system environment.
[PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection because it loads instructions from external markdown files and interpolates them into subagent prompts without validation.
Ingestion points: Agent persona definitions are read from the project-local ./agents/ directory and the global user directory ~/.claude/agents/.
Boundary markers: The instructions do not define specific delimiters or instructions to ignore embedded commands within the persona files before they are passed to subagents.
Capability inventory: The skill utilizes the Agent tool with subagent_type: 'general-purpose' to execute the loaded personas.
Sanitization: The markdown content of the agent files is not sanitized or filtered before processing.

team-builder