The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the Bash tool to execute tasks identified during the orchestration process. This is performed within a defined pipeline and only after the user has reviewed and approved the proposed execution plan.
[PROMPT_INJECTION]: The skill ingests user-provided task descriptions to determine execution logic and generate prompts for sub-agents. While this creates a surface for indirect prompt injection where the input data could attempt to override agent instructions, the skill implements a mandatory human-in-the-loop confirmation phase to mitigate this risk.

agent-orchestrate