The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the platform's Agent Teams API (including TeamCreate and TaskCreate) to spawn and manage specialized subagents. It specifically instructs the primary agent to use the Task tool with run_in_background: true to enable parallel processing in separate tmux panes.- [PROMPT_INJECTION]: The instructions direct the primary agent to forgo its default analysis behaviors in favor of immediate delegation to a 'planner' subagent. This is a functional requirement of the orchestration workflow.- [EXTERNAL_DOWNLOADS]: The skill references several external resources and tools, including the codex CLI and security skills like CodeGuard. It also suggests the use of various domain-specific subagents from the vendor 'sumik'.- [PROMPT_INJECTION]: The skill identifies a surface for indirect prompt injection where user input is interpolated into subagent prompts.\n
Ingestion points: User requests are captured and passed into the prompt parameter of the Task tool (Step 2 in WORKFLOW-GUIDE.md).\n
Boundary markers: Employs Markdown formatting as markers, though it lacks robust 'ignore instructions' delimiters for the user-supplied content.\n
Capability inventory: Subagents have the capability to execute shell commands and modify files in the background.\n
Sanitization: There is no documented logic for sanitizing or escaping the user's input before it is passed to the subagents.

orchestrating-teams