Google Jules

You can automate and manage Google Jules development workflows with this MCP server. It connects Jules for task creation, code review automation, repository handling, and sessionful browser-based interactions, enabling streamlined AI-assisted development directly from your tools.

How to use

You interact with the MCP server through an MCP client or integration that sends task, code, and session commands. Use it to create Jules tasks, monitor progress, send instructions, approve plans, and resume paused work. You can also analyze code changes, navigate repositories, manage branches, and run automated code review workflows. When you need to persist data or share context across sessions, rely on the built-in local storage for tasks and history. For remote deployments, you can run the server in a cloud or cloud-backed browser session mode to avoid keeping a local browser instance.

How to install

Prerequisites you need before running the MCP server are Node.js 18 or newer, TypeScript tooling, and Git access to any repositories you plan to manage.

Step-by-step setup: start by cloning the project, install dependencies, build, and test to ensure everything runs correctly.

# Clone the project
git clone https://github.com/samihalawa/google-jules-mcp.git
cd google-jules-mcp

# Install dependencies
npm install

# Build the project
npm run build

# Run tests
npm test

Configuration and usage notes

Configure your environment to control how the MCP server operates, including how the browser automation runs, how authentication is managed, and where data is stored. You can choose a session mode to fit your setup, such as fresh sessions for testing, or cookies or a Chrome profile for persistent authentication.

Key environment options you’ll typically configure include enabling headless mode, setting a connection timeout, and toggling debug logging. For remote deployments or cloud hosting, you can switch to a Browserbase-based session with the appropriate API keys and IDs. For manual login, you can provide Google authentication cookies and a cookies path for persistence.

Security and session management

The MCP relies on browser session data for Google authentication. You can run in several session modes such as a fresh session, a Chrome profile, cookie-based authentication, persistent browser data, or a remote Browserbase session for cloud deployments. Ensure you only provide cookies from legitimate sessions and protect your cookie data as sensitive credentials.

Troubleshooting

If browser automation fails, verify you can access jules.google.com and that you are logged into the correct Google account. Running with headless mode disabled can help you observe what happens during automation. For task creation issues, confirm repository names follow owner/repo-name format and that you have access to those repositories. If you encounter permission errors, check your data storage path permissions and repository access in your version control system.

Enable debug mode to get more insight during troubleshooting. You can start the server with debug output and visible browser interactions to capture screenshots and logs for analysis.

Development notes

This MCP uses browser automation to interact with Jules and persists task data locally. It is designed to work both locally and in cloud environments, with a focus on secure handling of browser sessions and avoiding hard-coded credentials.

Available tools

jules_create_task

Create a new Jules task with repository and description

jules_get_task

Retrieve detailed information about a specific Jules task

jules_send_message

Send messages or instructions to Jules within an active task

jules_approve_plan

Approve Jules execution plans for tasks

jules_resume_task

Resume paused or interrupted Jules tasks

jules_list_tasks

List tasks with filtering options and status probes

jules_analyze_code

Analyze code changes and project structure for automated insights

jules_bulk_create_tasks

Create multiple Jules tasks from a list of descriptions