deep-agents-orchestration

You can wire SubAgentMiddleware, TodoListMiddleware, and HumanInTheLoopMiddleware together to orchestrate subagents, plan multi-step tasks, and require explicit approval gates for sensitive operations within a single Deep Agents configuration.

Human-in-the-loop orchestration requires explicit human approval before sensitive operations execute. You need it for automated deployments or research automation where delegated subagent tasks carry risk and require an approval gate before proceeding.

Configure a deep agent by including a subagent for delegated specialized work, TodoListMiddleware to plan and track multi-step tasks, and HITL gates to handle sensitive operations, creating a self-contained configuration for complex automation.

Yes, subagent orchestration supports research automation by delegating specialized tasks to dedicated subagents, using TodoListMiddleware to track progress, and applying HumanInTheLoopMiddleware to ensure safe, scalable automation with approval gates.

No external dependencies are required. This orchestration provides a self-contained configuration that wires together the necessary middleware for subagent delegation, todo list planning, and human-in-the-loop approval gates.