GitHub Universe 2023 is underway in San Francisco and the developer centric platform has announced a slew of new AI-powered features for developers. At the forefront of these is GitHub Copilot Chat, a GPT-4 powered LLM (Large Language Model) that leverages GitHub’s massive code library to give you an AI assistant on steroids. GitHub Copilot Enterprise will be an offering for companies starting off at USD 39 per developer and it will be a more personalised AI assistant that’s in tune with each developers’ codebase to help provide more contextual solutions. Aside from these two major announcements, GitHub also unveiled new AI-powered security features to help developers avoid security pitfalls by suggesting fixes for what could be termed vulnerable code. And the last major announcement was GitHub Copilot Workspace which is a platform to allow developers to implement major changes across the entire codebase without having things falling though the cracks.
Thomas Dohmke, CEO, GitHub, said: “In March, we shared our vision of a new future of software development with Copilot X, where AI infuses every step of the developer lifecycle. Since then, we’ve been working to scale and mature the underlying technology–and in the process we created something even bigger. Our vision has manifested itself into a new reality for the world’s developers.
“Just as GitHub was founded on Git, today we are re-founded on Copilot. Open source and Git have fundamentally transformed how we build software. It is now evident that AI is ushering in the same sweeping change, and at an exponential pace. In just a short period, GitHub Copilot has expanded and evolved GitHub into the world’s leading AI-powered developer platform.
“We are certain this foundational transformation of the GitHub platform, and categorically new way of software development, is necessary in a world dependent on software. Every day, the world’s developers balance an unsustainable demand to both modernise the legacy code of yesterday and build our digital tomorrow. It is our guiding conviction to make it easier for developers to do it all, from the creative spark to the commit, pull request, code review, and deploy—and to do it all with GitHub Copilot deeply integrated into the developer experience,” Dohmke added.
These are just some of the many new additions that were announced at GitHub Universe 2023 to what can be called as the world’s biggest developer platform. Let’s look at these announcements in greater detail.
Yes, we know, an AI-assistant developer on steroids is as good as a proper developer. But like all AI-powered tools, GitHub Copilot Chat is much more powerful. It’s powered by GPT-4 so you’re getting all the rich natural language processing capabilities that OpenAI has incorporated into ChatGPT. So it can be put to work for a myriad of tasks such as writing unit tests for individual functions or modules, debugging code and a lot more. What’s even better is that it is not limited to just one window or terminal. GitHub Copilot Chat reads through your existing code and can also access the content that’s currently present in any open terminal and even the debugger. It uses all these pieces of information to provide a more contextual solution to whatever problems that the developer may be facing.
GitHub Copilot Chat is built in true developer-centric nature. It not only keeps the users’ code in mind as context but it can also detect common security vulnerabilities that might arise when writing code in a certain manner. Think of it as if there’s a constant code-review process running in the background at all times. Also, you can issue commands using shortcuts using slash commands and context variables. So generating a new set of unit tests is as simple as typing /tests and asking Copilot chat for a fix can happen just by typing /fix. Another helpful change is that ability to iterate on code inline. So you don’t have to continuously generate the entire code over and over again to incorporate simple changes.
Copilot Chat will be integrated into github.com as well as GitHub’s mobile app. This way, developers can play around with code, issue pull requests, work on documentation and ask questions in general using the app or the web interface. Moreover, the Advanced Code Search on GitHub can be used to keep tabs on the latest changes to some of the biggest and most popular open-source projects hosted on GitHub.
GitHub Copilot Chat will be available to the general public in December 2023 as part of their existing GitHub Copilot subscription. And like before, verified teachers and students as well as maintainers of popular projects will be given access to the tool for free. If you’re looking for an IDE integration then Copilot Chat is currently available as a preview within Jetbrains Suite starting today.
Built for enterprises that like to keep their codebases private and preferring a little bit of customisation, GitHub announced Copilot Enterprise starting at USD 39 per user per month. The new service will be available for enterprises starting February 2024. The enterprise offering will also bring the full feature set of the newly announced Copilot Chat.
Large Language Models inherently bring the experience of multiple seasoned developers to the table and this applies to not only generating code but also generating code without typical insecure coding patterns. GitHub Advanced Security proactively intercepts insecure coding patterns as they emerge, bolstering the security of Copilot’s code recommendations. The algorithm is finely tuned to spot and block prevalent security flaws such as hard-coded credentials, SQL injections, and path injections with real-time efficiency.
Moreover, GitHub Copilot Chat extends this protective envelope into integrated development environments (IDEs). It leverages its natural language prowess to not only detect security weaknesses but also to elucidate the underlying principles of the vulnerabilities. It goes a step further by proposing targeted remediations for problematic code segments. This capability is a game-changer, redefining the concept of ‘shift left’ in security paradigms.
One of the standout features is the new ‘code scanning autofix’ function, which leverages CodeQL to autonomously generate solutions for issues in JavaScript and TypeScript within pull requests. This accelerates the resolution process for developers, curtailing the risk of new vulnerabilities seeping into codebases.
The introduction of AI-powered secret scanning for detecting generic secrets, along with GitHub’s regular expression generator for crafting custom patterns makes it easier to detect compromised or leaked secrets with fewer false positive rates.
Navigating from concept to code remains a substantial challenge for developers, often beginning with the initial blueprint that transmutes an abstract idea into executable code through to the finalisation of a pull request. GitHub issues frequently serve as the birthplace for both new features and bug reports. To bridge this gap, GitHub Next’s research team has harnessed GPT-4’s analytical prowess to create the GitHub Copilot Workspace—an AI-augmented environment designed to facilitate the coding process.
Within Copilot Workspace, when a developer opens an issue, the system automatically generates a tentative plan for implementing the required changes. The workspace is dynamic and fully customisable, allowing developers to guide the AI, leveraging its comprehensive grasp of the issue’s context and the nuances of the codebase. The workspace aims to streamline the build, run, and test cycle, offering to autocorrect errors when they arise.
Effectively, Copilot Workspace operates akin to a co-pilot in a pair programming scenario—one that’s intimately familiar with the project’s every detail and capable of executing broad changes across the repository, from issue to pull request, all under the user’s directive.
Copilot Workspace, expected in 2024, seems like having an AI-powered Engineering Manager on board with your project. We expect that more details will emerge closer to launch.
