Deep Agents: The Batteries-Included Agent Harness for Complex AI Tasks

Introduction

Deep Agents is an opinionated, batteries-included agent harness built on top of LangChain and LangGraph. It provides a ready-to-run agent out of the box, eliminating the need to wire up prompts, tools, and context management manually. This framework is designed to handle complex agentic tasks by equipping agents with powerful capabilities.

Key features included:

• Planning: write_todos for task breakdown and progress tracking.
• Filesystem: read_file, write_file, edit_file, ls, glob, grep for reading and writing context.
• Shell access: execute for running commands, with sandboxing.
• Sub-agents: task for delegating work with isolated context windows.
• Smart defaults: Prompts that teach the model how to use these tools effectively.
• Context management: Auto-summarization for long conversations, large outputs saved to files.

Installation

To get started with Deep Agents, you can install it using pip or uv:

pip install deepagents
# or
uv add deepagents

Examples

Once installed, you can quickly create and invoke a Deep Agent:

from deepagents import create_deep_agent

agent = create_deep_agent()
result = agent.invoke({"messages": [{"role": "user", "content": "Research LangGraph and write a summary"}]})

Deep Agents are highly customizable. You can add your own tools, swap models, customize prompts, and configure sub-agents:

from langchain.chat_models import init_chat_model

agent = create_deep_agent(
    model=init_chat_model("openai:gpt-4o"),
    tools=[my_custom_tool],
    system_prompt="You are a research assistant.",
)

Why use Deep Agents

Deep Agents offers several compelling reasons for its adoption:

• 100% open source: MIT licensed and fully extensible.
• Provider agnostic: Works with Claude, OpenAI, Google, or any LangChain-compatible model.
• Built on LangGraph: Provides a production-ready runtime with streaming, persistence, and checkpointing.
• Batteries included: Planning, file access, sub-agents, and context management work out of the box.
• Get started in seconds: Install with pip install deepagents or uv add deepagents for a working agent.
• Customize in minutes: Easily add tools, swap models, and tune prompts as needed.

Links

• GitHub Repository
• Official Documentation
• API Reference
• LangChain Forum