Deep Agents Orchestration

This covers Deep Agents' three orchestration patterns: delegating work to specialized subagents via the task tool, planning complex workflows with todo lists, and requiring human approval for sensitive operations. The subagent delegation is especially useful when you need to isolate complex work or use specialized tools without cluttering your main agent's context. The examples show practical patterns like creating researcher subagents with custom tools, setting up approval workflows for database operations, and managing multi-step API development tasks. One thing to watch out for is that custom subagents don't inherit skills from the main agent, so you need to explicitly configure their capabilities.

skills.sh installs: 16.8K
GitHub stars: 1.3K
last updated: Aug 21, 2026

View on GitHub

Install

$ npx -y skills add langchain-ai/langchain-skills --skill deep-agents-orchestration --agent claude-code

Installs into .claude/skills of the current project.

Files

Deep Agents include three orchestration capabilities:

SubAgentMiddleware: Delegate work via task tool to specialized agents
TodoListMiddleware: Plan and track tasks via write_todos tool
HumanInTheLoopMiddleware: Require approval before sensitive operations

All three are automatically included in create_deep_agent().

Subagents (Task Delegation)

Use Subagents When	Use Main Agent When
Task needs specialized tools	General-purpose tools sufficient
Want to isolate complex work	Single-step operation
Need clean context for main agent	Context bloat acceptable

Main agent has `task` tool -> creates fresh subagent -> subagent executes autonomously -> returns final report.

Default subagent: "general-purpose" - automatically available with same tools/config as main agent.

Create a custom "researcher" subagent with specialized tools for academic paper search.

SKILL.md on GitHub