GitHub Copilot app adds slash commands for faster workflows

GitHub has expanded its Copilot desktop app with six key slash commands designed to streamline different phases of the development workflow. Each command addresses a specific need and can be discovered by typing / in the chat composer, where an autocomplete menu reveals available options.

The /plan command helps developers break down tasks before writing code. It switches the session into Plan mode and is useful for planning new features (identifying files, components, and dependencies), preparing for large refactors (mapping out complex changes and uncovering risks), and triaging bugs (exploring possible causes and outlining diagnostic steps). For example, a developer can type "/plan I need to add two-factor authentication to our application. Help me break down the work involved, identify what files need to change, and outline an implementation approach" to get structured guidance before implementation begins.

The /spar command acts as a devil's advocate, pressure-testing ideas by having Copilot question assumptions and point out potential risks or trade-offs. It can be used to validate architecture choices (e.g., using Redis for caching), compare implementation options (REST versus GraphQL), review migration plans, or challenge performance optimizations. A developer might use it like "/spar I'm planning to use Redis as a caching layer for our product API. Challenge my approach and point out any scalability or consistency concerns I may have missed."

Once a plan is ready, /autopilot takes over to handle implementation. It switches the session into Autopilot mode and helps developers work through implementation steps and iterate. It can be used to implement new features by handing off a task and letting Copilot work through the steps, or to complete larger maintenance tasks like updating dependencies or refactoring code.

The /rubber-duck command provides a second opinion by using a different model to independently review work. It surfaces blind spots, questions assumptions, and catches issues the primary model may have missed—particularly valuable for complex refactors, architectural decisions, and migration plans. Unlike the primary model, this fresh perspective helps validate work before a pull request is opened or a deployment moves forward.

The /create-canvas command turns conversations into interactive experiences. Instead of scrolling through chat, developers can create visualizations, dashboards, or custom workflows they can interact with—such as interactive diagrams showing service connections or issue triage boards for categorizing open issues.

Finally, /orchestrate coordinates work across multiple repositories. It breaks larger efforts into smaller tasks that can move forward in parallel, helping developers track and coordinate updates when a feature touches multiple codebases or when larger projects require work across implementation, testing, documentation, and deployment. The article notes that users do not need to memorize commands; typing / and exploring the autocomplete menu is sufficient to get started.