deepagent-lab 0.1.3

JupyterLab extension with chat interface for DeepAgents

deepagent-lab

A JupyterLab extension providing an AI agent chat interface with notebook manipulation capabilities and human-in-the-loop controls.

Features

• Chat Interface: Sidebar for natural conversations with your agent
• Notebook Manipulation: Built-in tools for creating, editing, and executing Jupyter notebooks
• Human-in-the-Loop: Review and approve agent actions before execution
• Context Awareness: Automatically sends workspace and file context to your agent
• Agent Portability: Use any other langgraph-compatible agent seamlessly

Installation

pip install deepagent-lab

Quick Start

1. Set up your environment (copy .env.example to .env and configure):

# Required: Jupyter server configuration (must match your JupyterLab startup)
DEEPAGENT_JUPYTER_SERVER_URL=http://localhost:8889
DEEPAGENT_JUPYTER_TOKEN=8e2121e58cd3f9e13fc05fc020955c6e # Generate with python3 -c "import secrets; print(secrets.token_hex(16))"

# If using the default agent, Anthropic API key is required
ANTHROPIC_API_KEY=your-api-key-here

# Or if you want to use your agent, specify the location here
DEEPAGENT_AGENT_SPEC=./my_agent.py:agent

2. Start JupyterLab with matching server URL and token:

# Start JupyterLab with values matching your .env file
jupyter lab --port=8889 --IdentityProvider.token=8e2121e58cd3f9e13fc05fc020955c6e

Important: The Jupyter server URL and token in your .env file must match the values JupyterLab uses when starting up. This allows the agent to connect to notebook kernels for code execution.

3. Open the chat interface by clicking the chat icon in the right sidebar

4. Start chatting with your agent!

Agent Configuration

The extension uses the DEEPAGENT_ prefix for all environment variables, enabling full compatibility with deepagent-dash.

Quick Configuration

Option 1: Use the default agent

• The extension includes a built-in agent for notebook manipulation
• No configuration needed - just start chatting!

Option 2: Use a custom agent

Create a custom agent and point to it using environment variables:

# Agent spec in format "module_or_file:variable"
DEEPAGENT_AGENT_SPEC=./my_agent.py:agent

Environment Variables

All configuration uses the DEEPAGENT_ prefix for compatibility with deepagent-dash:

Variable	Purpose	Default
DEEPAGENT_AGENT_SPEC	Agent location (path:variable)	Uses default agent
DEEPAGENT_WORKSPACE_ROOT	Working directory for agent	JupyterLab root
DEEPAGENT_MODEL_NAME	Model identifier	anthropic:claude-sonnet-4-20250514
DEEPAGENT_MODEL_TEMPERATURE	Model temperature (0.0-1.0)	0.0
DEEPAGENT_JUPYTER_SERVER_URL	Jupyter server URL	http://localhost:8889
DEEPAGENT_JUPYTER_TOKEN	Jupyter auth token	12345
DEEPAGENT_VIRTUAL_MODE	Safe mode for filesystem	true
DEEPAGENT_DEBUG	Enable debug logging	false

See .env.example for complete configuration options.

Creating Custom Agents

from deepagents import create_deep_agent
from deepagents.backends import FilesystemBackend
from langgraph.checkpoint.memory import MemorySaver
import os

# Agent discovers workspace automatically
workspace = os.getenv('DEEPAGENT_WORKSPACE_ROOT', '.')

agent = create_deep_agent(
    model="anthropic:claude-sonnet-4-20250514",
    backend=FilesystemBackend(root_dir=workspace, virtual_mode=True),
    checkpointer=MemorySaver(),
    tools=[...your_tools...]
)

Save this as my_agent.py and configure:

DEEPAGENT_AGENT_SPEC=./my_agent.py:agent

Agent Portability

Agents configured for deepagent-lab work seamlessly in deepagent-dash:

# Same .env file works for both!
DEEPAGENT_AGENT_SPEC=./my_agent.py:agent
DEEPAGENT_WORKSPACE_ROOT=/path/to/project

# Run in JupyterLab
jupyter lab

# Or run in Dash
deepagent-dash run

Interface Controls

• ⟳ Reload: Reload your agent without restarting JupyterLab
• Clear: Start a new conversation thread
• Status Indicator:
  • 🟢 Green: Agent ready
  • 🟠 Orange: Agent loading
  • 🔴 Red: Agent error

Development

See CONTRIBUTING.md for development setup and guidelines.

License

MIT License - see LICENSE for details.