Jupyter Ai Agents
🪐 🤖 AI Agents for JupyterLab with 🔧 MCP tools - Chat interface for optimized notebook interaction and code execution.
Last verified:
What is Jupyter Ai Agents?
Jupyter AI Agents is a flexible and powerful solution for using and creating AI Agents to interact with Jupyter Notebooks. It empowers AI models to interact with and modify Jupyter Notebooks comprehensively, equipped with tools like insert_cell, execute_cell, adding code cells, inserting markdown cells, and executing code. The agent operates on the entire Notebook, not just at the cell level, enabling more comprehensive and seamless modifications.
Key features include Dynamic Notebook Interaction (adding, editing, and executing cells dynamically based on user instructions or notebook events), Remote Agent Execution (running independently on a different kernel), Event-Based Modifications (reacting to Kernel and Notebook changes in real time), Advanced Jupyter Tools (inserting markdown or code cells, running code), and Built on RTC (Real-Time Collaboration ensuring seamless communication). The library uses the LangChain Agent Framework to manage interactions between AI models and tools.
Jupyter AI Agents is designed for data scientists, ML engineers, developers, and researchers who use Jupyter Notebooks and want to automate repetitive tasks, generate code from natural language, debug errors, analyze data, create visualizations, and enhance productivity. It supports multiple AI model providers including Azure OpenAI, GitHub Copilot, OpenAI, and Anthropic Claude models. Available agents include the Prompt Agent (generates code/markdown cells from natural language) and Explain Error Agent (explains errors with actionable solutions).
The tool is available as an open-source Python library (pip install jupyter_ai_agents) and can be used in JupyterLab with the Collaboration extension for real-time visibility of modifications. It supports multiple deployment modes (In-Kernel, Out-Kernel Stateless, Out-Kernel Stateful) and interaction modes (Ask Mode, Listen Mode).
Jupyter Ai Agents pricing
Pricing model: Freemium
Jupyter AI Agents is an open-source Python library available via pip (pip install jupyter_ai_agents) at no cost. The core library is free to use. Users need to provide their own AI model provider access (Azure OpenAI, GitHub Copilot, OpenAI, or Anthropic) which may have separate pricing. The tool can also be deployed with Datalayer online services which may have their own pricing for cloud hosting and remote runtimes.
Jupyter Ai Agents pros
- Operates on entire notebook, not just cell level
- Dynamic cell insertion, editing, and execution
- Real-time modifications visible in JupyterLab via RTC
- Remote agent execution on separate kernel
- Event-based automatic modifications without user intervention
- Supports multiple AI model providers (Azure OpenAI, GitHub Copilot, OpenAI, Anthropic)
- Prompt Agent generates code from natural language input
- Explain Error Agent provides actionable debugging solutions
- Can create custom agents by extending base class
- LangChain Agent Framework for robust agent management
- Multiple deployment modes for flexibility
- Ask Mode and Listen Mode interaction options
- Open-source library with GitHub repository
- Simple pip installation process
- Automates repetitive data analysis and visualization tasks
Jupyter Ai Agents cons
- No native UI yet - primarily CLI-based interaction
- Custom agent creation guide under construction
- In-Kernel mode not recommended for production
- Stateless mode lacks Agent Memory features
- Requires JupyterLab with Collaboration extension for real-time view
- Modify Code Cell and Modify Markdown Cell tools still being implemented
- Limited to JupyterLab and Datalayer Platform currently
- Requires separate AI model provider API access and configuration
Frequently asked questions about Jupyter Ai Agents
What is Jupyter AI Agents?
Jupyter AI Agents is a flexible and powerful solution for using and creating AI Agents to interact with Jupyter Notebooks. It empowers AI models to interact with and modify Jupyter Notebooks using tools like insert_cell, execute_cell, adding code cells, inserting markdown cells, and executing code. The agent operates on the entire notebook, not just at the cell level.
What AI model providers are supported?
Jupyter AI Agents supports multiple AI model providers including Azure OpenAI, GitHub Copilot, OpenAI, and Anthropic Claude models. You can configure the model provider and model name when launching the agent via CLI.
How do I install Jupyter AI Agents?
To install Jupyter AI Agents, run 'pip install jupyter_ai_agents'. Alternatively, you can clone the repository and install from source using 'git clone https://github.com/datalayer/jupyter-ai-agents', then 'cd jupyter-ai-agents' and 'pip install -e .'.
What is the Prompt Agent?
The Prompt Agent generates code cells based on natural language input. It leverages AI models to generate code, markdown, add new cells, and modify notebook content. Use cases include code generation, markdown generation, and notebook modification based on existing cells.
What is the Explain Error Agent?
The Explain Error Agent explains errors encountered in a notebook. It leverages AI models to explain the error in a code cell, providing insights into the error and potential solutions by adding new cells with fixes or explanations.
What are the deployment modes?
There are three deployment modes: In-Kernel (agent requested directly in Jupyter Kernel, for development only), Out-Kernel Stateless (requested via CLI, no Agent Memory), and Out-Kernel Stateful (separate process via REST endpoints, with Agent Memory features).
What are Ask Mode and Listen Mode?
Ask Mode is when the Agent is explicitly requested by the user. Listen Mode is when the Agent observes Notebook and Kernel events and is requested automatically when a specific event occurs without user action.
What tools are currently available?
Current tools include: Add and Execute Code Cell, Add Markdown Cell, Insert and Execute Code Cell at specific position, and Insert Markdown Cell at specific position. Tools like Modify Code Cell and Modify Markdown Cell are being implemented.
Can I create custom agents?
Yes, you can create custom Jupyter AI Agents like Code Refactor Agent or Documentation Agent by extending the base class. The LangChain Agent Framework manages interactions between the AI model and tools. A guide for creating custom agents is under construction.
What use cases does Jupyter AI Agents support?
Supported use cases include Automated Data Analysis (generating and executing Python scripts for data analysis, visualizations, preprocessing), Error Debugging (identifying and explaining errors with actionable suggestions). Future agents will support code refactoring and improved documentation.