GitHub Copilot app: how to write your first prompt

GitHub published guidance for developers beginning to use the GitHub Copilot app, a tool that assists with coding tasks through conversational prompts. The article walks through the key steps to get started and clarifies that users do not need to write perfect prompts from the outset.

To begin, users connect the app to a project or local folder, giving Copilot the context and code it needs to work on the task. From the GitHub Copilot app home screen, users can select a project they have worked with before or add a new one; local folders are also an option for code that exists only on the user's computer. Once the project is selected, the user is ready to write their first prompt.

The article emphasizes simplicity in prompt writing. Users describe what they want to accomplish in plain English—for example, 'Add a most-funded sort option to the games list'—without needing to learn special syntax or perfect phrasing. Copilot can examine the project, find relevant parts of the codebase, and work on the change. If the first attempt does not match expectations, users can provide additional details or ask for adjustments; the guide describes prompting as an iterative process where you do not have to anticipate every detail before beginning. The best prompt, the article states, is often the one that gets you moving.

The app also lets users choose which AI model handles their task, recognizing that different models have different strengths—some are better suited to complex reasoning while others handle simpler tasks more quickly. For users just getting started, the default model is a good place to begin, and users do not need to understand every difference between available models before using the app. As work progresses, users can switch models when a task requires more complex reasoning or when the first approach is not giving the results they need. The article frames the model selector as a tool to reach for when the task calls for it, rather than something that must be configured before every session.

Beyond text, the GitHub Copilot app includes built-in voice input. Users can speak to describe their task, and the app converts speech into text in the prompt box, allowing review and editing before sending. This can be useful when thinking through a problem out loud or when a longer request is easier to describe verbally than type.

Users can also customize how their session handles work. From the session title, users can select a different agent or enable remote control for the session. Different agents can be configured for different types of work, and remote sessions allow work to run on the web in addition to locally—users can start a task, close their laptop, and return to the session later from another device without losing their place. The article notes that these options are not things that must be configured before the first prompt; they are there when users need more control over how a task is handled.

The guide concludes by encouraging users to start small and iterate. The first prompt does not need to be perfect; users should pick a project they already know, describe the desired change in plain language, and observe what happens. From there, they can refine the request, try a different model, or adjust how the session runs. The article states that the more users use the app, the more natural these choices become, and the important part is getting started.