The canvases in the GitHub Copilot application are customizable interfaces that take shape according to how the user works. The user and the agent can work on the same canvas. A canvas can be created in the following forms:
- Kanban board
- Issue triage board
- Release checklist
- Dashboard
- Form
- Spreadsheet
Canvases work both ways: The agent can update content, while the user can make changes using buttons, cards, and filters. No coding or design work is required to create one. It is enough to open an agent session, enter the `/create-canvas` command, and describe the request in plain English. The description should include the following three elements:
- The workflow the canvas will support
- The actions that can be performed in the interface
- What the agent can do
The agent creates the interface based on the description and opens it in the right panel. The first version can later be improved; the user can ask for a new column or filter to be added, open pull requests to be retrieved, or the canvas to be turned into a daily checklist. Although there is no fixed layout menu, workflows that can be defined can be turned into canvases.
The created canvas is saved as an extension; it can be shared with the team along with the project or saved as a personal extension. When the user changes a field or moves a card, the shared state is instantly reflected in the agent. The agent can also add content or move cards using the same actions on the canvas.
The Awesome Copilot community has ready-made canvas extensions. Users can install one of them and customize it according to their own workflow. To get started, a simple board or checklist can be defined with `/create-canvas`.
Why it matters
This approach moves Copilot beyond being merely a chat space for receiving suggestions and responses, turning it into a workspace where teams can track shared workflows. Particularly for tasks such as issue triage, release preparation, and daily checks, having the user and the agent work from the same state information provides a shared working arrangement instead of manual handoffs. The shareable extension model makes it possible for the prepared interfaces not to remain limited to a single user’s personal setup and to be shared across the team. However, the lack of a fixed layout menu makes the interface’s functionality dependent on how clearly the workflow is defined; the open question is to what extent different teams can use this flexibility consistently.
Background
Copilot is not a new name in the FikirPilot archive: we have published 5 stories mentioning the name in the last 90 days; the most recent was dated September 25, 2026.
Term: agent
An artificial intelligence agent is software that calls tools and carries out multi-step tasks to achieve a goal rather than producing a single response.