Jules AI: Google’s Autonomous Coder Gets a Massive Upgrade
Google appears to be revamping its suite of AI lab products, and its autonomous coding agent, Jules, is at the forefront with major new upgrades. This makes the already powerful tool even more efficient and capable. For those unfamiliar, Jules is Google’s autonomous coding assistant, completely free to access, designed to help developers fix bugs, add documentation, and build new features with ease.
[0:15.867] [Jules UI showing the tool’s name and features.]
Jules is a smart coding assistant that seamlessly integrates directly with GitHub. It understands your entire codebase and works asynchronously, meaning you can assign a task and move on to other work while Jules handles the coding in the background. Now, with a wave of new updates, its capabilities have expanded significantly.
[0:39.117] [An overview graphic titled ‘Jules Updates’ showing various new features like Jules CLI, Web Search, Code Review, and Memory.]
The latest updates bring a host of powerful new features. This includes integration through a new Jules API, a brand-new memory system for enhanced context retention and long-term project awareness, and a command-line interface, alongside numerous bug fixes and performance upgrades. Let’s dive into what makes these updates a game-changer for developers.
[0:43.767] [An animation showing the Jules octopus icon connected to GitHub, VS Code, and a terminal, under the title ‘Changelog’.]
With the introduction of the Jules API, you can now integrate Jules directly into your own systems, applications, or CI/CD pipelines. This opens up a new level of automation and customization for your development workflow.
[0:50.817] [An animation of a purple octopus character typing a command into a terminal.]
You can now create remote tasks directly from your terminal, manage project awareness over the long term, and securely handle credentials like API keys and passwords, making the entire process more robust and integrated.
# Create a remote task
jules remote new \
--repo torvalds/linux \
--session "write u_"
[1:02.817] [A screenshot of the ‘Introducing the Jules API’ documentation, showing a code snippet for sending a request.]
Google has officially launched the Jules API, a significant step forward for the autonomous coding agent. This new API allows developers to programmatically access Jules’s capabilities, unlocking deep automation and powerful integrations across the entire software development lifecycle.
You can now programmatically access Jules’s capabilities to automate your work and build powerful integrations. The Jules API is designed to help you seamlessly integrate Jules into your existing development workflows, unlocking new ways to automate and enhance the entire software development lifecycle.
// Send Request to Jules API
core.info('Sending request to Jules API...');
const response: AxiosResponse<any> = await axios.post(endpoint, body);
// Handle the Response
core.info('Session created successfully.');
core.setOutput('session-name', response.data.name);
} catch (error) {
core.setFailed(`Action failed with error: ${error.message}`);
}
[1:22.957] [A terminal window showing a prompt being sent to Jules to write a blog post.]
With the Jules API, you can build custom integrations tailored to your needs. For instance, you could connect Jules with Slack to create chatbots that handle coding tasks directly within your communication channels. Simply dispatch a task, and Jules will create a plan, execute it, and make changes to the repository, all while you track its progress via a unique session ID.
> ask jules to write a blog post about hamsters in this starter blog
[1:38.907] [A chat interface where a user is asking Jules to create a new blog post with specific frontmatter.]
The possibilities extend to automated bug fixing and feature implementation by linking Jules to project management tools like Linear or Jira. You can even embed Jules directly into your CI/CD pipelines using services such as GitHub Actions, creating a truly automated development environment.
[1:52.227] [A screenshot of the Jules API documentation explaining authentication and API concepts.]
Google’s documentation provides a quickstart guide, demonstrating how straightforward it is to spin up a new task, or “session,” using a simple curl command. This makes it incredibly easy to get Jules working on a new project or feature request with just a single API call, marking a huge leap forward for integrating AI into modern software development.
[2:26.337] [A screenshot of the ‘Jules in the command line’ documentation, showing the Jules CLI interface.]
For developers who live in the terminal, Google has introduced the Jules CLI. This powerful new command-line interface gives you direct control over the AI coding agent right from your console.
[2:32.417] [A terminal window showing the installation of the Jules CLI via npm.]
With the CLI, you can create and manage tasks, list active sessions, and even apply patches locally to test code changes instantly before committing them to GitHub. It’s fully scriptable and composable, allowing for deeper automation with tools like gh and jq. The CLI also includes an interactive terminal dashboard for a more guided task management experience.
[2:55.337] [A slide listing the key features of the Jules CLI: Direct Control, Apply Patches Locally, Scriptable & Composable, and Interactive Dashboard.]
Key Features:
- Direct Control: Create tasks (
jules remote new), list active sessions (jules remote list), and monitor Jules without leaving your command line.- Apply Patches Locally: Instantly pull work-in-progress code from an active Jules session and apply it to your local machine.
- Scriptable & Composable: Integrate Jules into your automations by piping in output from other tools like
gh,jq, orcat.- Interactive Dashboard: For a more guided experience, launch the built-in terminal user interface (TUI) to create and manage tasks step-by-step.
[3:06.667] [A slide showing how to install the Jules CLI globally with npm or run it directly with npx, along with some starter commands.]
Installation is simple. You can install it globally via npm or run it directly without a permanent installation using npx. This update makes working with your autonomous coding agent faster and more flexible, integrating perfectly into your existing development workflow.
# How to Install:
Install globally via npm: npm install -g @google/jules
Or run directly without a permanent installation: npx @google/jules
# Starter Commands to Try:
See all available commands: jules help
List all repos connected to Jules: jules remote list --repo
Create a new task in a specific repo: jules remote new --repo torvalds/linux --session "write unit tests"
[3:18.117] [An animation under the title ‘File Selector’ showing a hand picking and arranging folder icons on a shelf.]
A new File Selector feature now gives developers precise control over which files the AI agent should modify. Instead of Jules scanning the entire codebase, you can now manually select specific files for it to edit, debug, or document, focusing its efforts exactly where they are needed.
[3:34.207] [A UI screenshot for ‘Use Environment Variables In Jules’, showing how to set up keys and secrets.]
Jules now supports environment variables, making the coding agent smarter and more adaptable to real-world projects. You can provide Jules with repository-level environment variables, giving it access to the project-specific configurations it needs to run builds, execute tests, or interact with external services.
[3:45.717] [A slide explaining ‘How It Works’ for environment variables: Add variables in repo settings, enable them for a task, and get task-long access.]
The process is simple: add variables in your repository settings, enable them when you start a new task, and Jules will retain access to them for the entire duration of that task. This feature unlocks new and more complex workflows, allowing for more sophisticated automation.
[4:42.507] [A screenshot of the ‘Memories’ feature UI within Jules, listing several saved context points about a project.]
One of the most exciting updates is Jules Memory for repositories. This groundbreaking feature allows the AI agent to learn from your interactions, saving your preferences, corrections, and workflow habits as you work.
[5:00.287] [A slide explaining the ‘Jules Memory for Repositories’ feature, its benefits, and settings.]
The next time you run the same or a similar task in that specific repository, Jules will reference its memory to better anticipate your needs and follow your established patterns, leading to more accurate results with less guidance.
This makes Jules a true long-term coding partner that improves over time. You can easily toggle this feature on or off in your repository settings, allowing the agent to grow with you and your projects.
[5:30.317] [A screenshot showing the image upload functionality within the Jules chat interface.]
To improve communication and context, Jules now supports image uploads. When creating a task, you can upload JPEG and PNG images (up to 5MB) to provide crucial visual context. This is perfect for showing frontend bugs, sharing UI mockups, or providing design inspiration, helping Jules better understand exactly what you need.
[5:57.067] [A four-panel graphic showing how Jules performs multiple tasks simultaneously.]
The review process has also been enhanced with a new stacked diff view. Previously, developers had to switch between tabs to see changes across different files. Now, the diff viewer displays all changes vertically on a single screen. This stacked layout provides better context, speeds up the review process, and makes it easier to understand the relationships between edits across your codebase.
[6:34.207] [A slide titled ‘Improved Jules Critic’ showing a running code review with detailed feedback.]
The Jules Critic agent has been significantly improved to provide more insightful and reliable code feedback. To increase transparency, the critic’s real-time thought process is now visible in the UI, offering a step-by-step evaluation as it analyzes your code. It also incorporates more contextual information to deliver more accurate and relevant feedback on potential bugs and logic flaws.
[7:08.537] [A tweet from the Jules agent account announcing a demo of Jules, Stitch, Gemini, and remote agents at GitHub Universe.]
Finally, Google is teasing an exciting integration between Jules and Stitch, its AI agent for UI and UX development. This upcoming connection will create a seamless workflow where you can design beautiful user interfaces in Stitch and then hand them off to Jules to automatically build the corresponding backend logic. This end-to-end, AI-powered development experience promises to revolutionize how applications are built.