Deep Agents From Scratch

🧱 Deep Agents from Scratch

[Deep Research](https://academy.langchain.com/courses/deep-research-with-langgraph) broke out as one of the first major agent use-cases along with coding. Now, we've seeing an emergence of general purpose agents that can be used for a wide range of tasks. For example, [Manus](https://manus.im/blog/Context-Engineering-for-AI-Agents-Lessons-from-Building-Manus) has gained significant attention and popularity for long-horizon tasks; the average Manus task uses ~50 tool calls!. As a second example, Claude Code is being used generally for tasks beyond coding. Careful review of the [context engineering patterns](https://docs.google.com/presentation/d/16aaXLu40GugY-kOpqDU4e-S0hD1FmHcNyF0rRRnb1OU/edit?slide=id.p#slide=id.p) across these popular "deep" agents shows some common approaches:

• Task planning (e.g., TODO), often with recitation
• Context offloading to file systems
• Context isolation through sub-agent delegation

This course will show how to implement these patterns from scratch using LangGraph!

🚀 Quickstart

Prerequisites

• Ensure you're using Python 3.11 or later.
• This version is required for optimal compatibility with LangGraph.

python3 --version

• uv package manager

curl -LsSf https://astral.sh/uv/install.sh | sh
# Update PATH to use the new uv version
export PATH="/Users/$USER/.local/bin:$PATH"

Installation

1. Clone the repository:

git clone https://github.com/langchain-ai/deep-agents-from-scratch.git
cd deep-agents-from-scratch

1. Install the package and dependencies (this automatically creates and manages the virtual environment):

uv sync

1. Create a .env file in the project root with your API keys:

# Create .env file
touch .env

Add your API