jupyter_ai_acp_client

A proof-of-concept package providing a client implementation of the Agent Client Protocol (ACP) in Jupyter AI v3, as well as helper classes for other developers to use when custom AI personas wrapping ACP agents.

This package provides a default ACP client implementation as JaiAcpClient. This client provides a prompt_and_reply() method which calls the ACP server and streams the reply back to the chat. In addition, it provides file read, file write, and terminal use capabilities.

This package also provides a default BaseAcpPersona class which can be easily extended to add ACP agents as AI personas in JupyterLab. This base class takes an additional executable argument which starts the ACP agent server. This package also provides a default ACP client implementation as JaiAcpClient.

• BaseAcpPersona automatically creates new subprocesses for the ACP agent and client when needed. These are stored as class attributes, so all instances of the same ACP persona share a common ACP agent subprocess.

• Since BaseAcpPersona inherits from BasePersona, subclasses can be provided simply as entry points to become available for use in Jupyter AI. (see documentation)

• Personas based on ACP now just need to derive from BaseAcpPersona and define the persona name, the persona avatar, and the executable starting the ACP agent server.

For example, the @Claude-ACP persona is defined in claude.py using less than 20 lines of code:

class ClaudeAcpPersona(BaseAcpPersona):
    def __init__(self, *args, **kwargs):
        executable = ["claude-code-acp"]
        super().__init__(*args, executable=executable, **kwargs)

    @property
    def defaults(self) -> PersonaDefaults:
        avatar_path = str(os.path.abspath(
            os.path.join(os.path.dirname(__file__), "..", "static", "claude.svg")
        ))

        return PersonaDefaults(
            name="Claude-ACP",
            description="Claude Code as an ACP agent persona.",
            avatar_path=avatar_path,
            system_prompt="unused"
        )

Currently, this package provides 3 personas:

1. @Test-ACP (a test persona that echoes responses)
2. @Claude-ACP
   • requires claude-code-acp, installed via npm install -g @zed-industries/claude-code-acp
3. @Kiro
   • requires kiro-cli, installed from https://kiro.dev

Dependencies

Required:

• JupyterLab >= 4.0.0
• jupyter-ai-persona-manager>=0.0.5
• agent_client_protocol

Optional

• claude-code-acp (enables @Claude-ACP)
• kiro-cli (enables @Kiro)

Install

To install the extension, execute:

pip install jupyter_ai_acp_client

Uninstall

To remove the extension, execute:

pip uninstall jupyter_ai_acp_client

Troubleshoot

If you are seeing the frontend extension, but it is not working, check that the server extension is enabled:

jupyter server extension list

If the server extension is installed and enabled, but you are not seeing the frontend extension, check the frontend extension is installed:

jupyter labextension list

Contributing

Development install

Note: You will need NodeJS to build the extension package.

The jlpm command is JupyterLab's pinned version of yarn that is installed with JupyterLab. You may use yarn or npm in lieu of jlpm below.

# Clone the repo to your local environment
# Change directory to the jupyter_ai_acp_client directory

# Set up a virtual environment and install package in development mode
python -m venv .venv
source .venv/bin/activate
pip install --editable ".[dev,test]"

# Link your development version of the extension with JupyterLab
jupyter labextension develop . --overwrite
# Server extension must be manually installed in develop mode
jupyter server extension enable jupyter_ai_acp_client

# Rebuild extension Typescript source after making changes
# IMPORTANT: Unlike the steps above which are performed only once, do this step
# every time you make a change.
jlpm build

You can watch the source directory and run JupyterLab at the same time in different terminals to watch for changes in the extension's source and automatically rebuild the extension.

# Watch the source directory in one terminal, automatically rebuilding when needed
jlpm watch
# Run JupyterLab in another terminal
jupyter lab

With the watch command running, every saved change will immediately be built locally and available in your running JupyterLab. Refresh JupyterLab to load the change in your browser (you may need to wait several seconds for the extension to be rebuilt).

By default, the jlpm build command generates the source maps for this extension to make it easier to debug using the browser dev tools. To also generate source maps for the JupyterLab core extensions, you can run the following command:

jupyter lab build --minimize=False

Development uninstall

# Server extension must be manually disabled in develop mode
jupyter server extension disable jupyter_ai_acp_client
pip uninstall jupyter_ai_acp_client

In development mode, you will also need to remove the symlink created by jupyter labextension develop command. To find its location, you can run jupyter labextension list to figure out where the labextensions folder is located. Then you can remove the symlink named @jupyter-ai/acp-client within that folder.

Testing the extension

Server tests

This extension is using Pytest for Python code testing.

Install test dependencies (needed only once):

pip install -e ".[test]"
# Each time you install the Python package, you need to restore the front-end extension link
jupyter labextension develop . --overwrite

To execute them, run:

pytest -vv -r ap --cov jupyter_ai_acp_client

Frontend tests

This extension is using Jest for JavaScript code testing.

To execute them, execute:

jlpm
jlpm test

Integration tests

This extension uses Playwright for the integration tests (aka user level tests). More precisely, the JupyterLab helper Galata is used to handle testing the extension in JupyterLab.

More information are provided within the ui-tests README.

AI Coding Assistant Support

This project includes an AGENTS.md file with coding standards and best practices for JupyterLab extension development. The file follows the AGENTS.md standard for cross-tool compatibility.

Compatible AI Tools

AGENTS.md works with AI coding assistants that support the standard, including Cursor, GitHub Copilot, Windsurf, Aider, and others. For a current list of compatible tools, see the AGENTS.md standard. This project also includes symlinks for tool-specific compatibility:

• CLAUDE.md → AGENTS.md (for Claude Code)

• GEMINI.md → AGENTS.md (for Gemini Code Assist)

Other conventions you might encounter:

• .cursorrules - Cursor's YAML/JSON format (Cursor also supports AGENTS.md natively)
• CONVENTIONS.md / CONTRIBUTING.md - For CodeConventions.ai and GitHub bots
• Project-specific rules in JetBrains AI Assistant settings

All tool-specific files should be symlinks to AGENTS.md as the single source of truth.

What's Included

The AGENTS.md file provides guidance on:

• Code quality rules and file-scoped validation commands
• Naming conventions for packages, plugins, and files
• Coding standards (TypeScript, Python)
• Development workflow and debugging
• Backend-frontend integration patterns (APIHandler, requestAPI(), routing)
• Common pitfalls and how to avoid them

Customization

You can edit AGENTS.md to add project-specific conventions or adjust guidelines to match your team's practices. The file uses plain Markdown with Do/Don't patterns and references to actual project files.

Note: AGENTS.md is living documentation. Update it when you change conventions, add dependencies, or discover new patterns. Include AGENTS.md updates in commits that modify workflows or coding standards.

Packaging the extension

See RELEASE