Pydantic AI Kernel

Pydantic AI Kernel is a wrapper around the pydantic-ai agent that enables interaction through the Jupyter kernel messaging protocol. It provides an easy way to create and share python based agents (with custom tools, ...).

Instead of building a custom CLI for a chatbot, this library leverages the Jupyter ecosystem to provide a free UI and a proven API for agent access. No CLI is implemented here - just the kernel.

This architecture allows for the spawning of multiple kernel instances via the Jupyter kernel gateway, enabling access to agents over web sockets.

Key Features

• No frontend implemented : just kernel. Frontend can be any jupyter frontend.
• Language-Agnostic Protocol: Based on Jupyter messaging.
• Streamed Output: streaming of text output
• Flexible Configuration: Configure system prompts, tools, mcp servers, models and inference providers (via pydantic-ai).
• Native web socket API : Jupyter kernel gateway
• Magics: Custom commands with TAB completion, for example : view history and tool calling, set up config, ...
• Interactive Tooling: Supports user validation for tools requiring approval.
• Different output types : kernel can send several data type for its output, supported by jupyter frontend (markdown, raw text, html, jupyter widgets, ...)

Getting Started

Installation

Install the package within a Python virtual environment:

pip install pydantic-ai-kernel

Usage

You can interact with the kernel through various Jupyter frontends:

• CLI: Install jupyter-console and run: bash jupyter console --kernel pydantic_ai
• Notebooks: Select pydantic_ai from the kernel list in your IDE.
• Jupyter Lab: Use the kernel directly within the Lab environment.
• Silik Signal Messaging: Access the kernel through the Signal Message Application (see here).
• jpterm : a fancy jupyter lab interface, in the terminal ! https://davidbrochart.github.io/jpterm/
• euporie : a fancy kernel frontend : https://euporie.readthedocs.io/en/latest/pages/installation.html

Jupyter Lite is not supported since the kernel is based on IPykernel, which is not accessible in Pyodide.

See Frontends for installation details and examples.

Quick Start

Once the kernel is running, use the %config magic to set up your agent:

%config

Refer to %help <magic_name> or %magic for detailed documentation on available commands.

You can also set up a configuration file manually : Configuration