Session M7-04: Build and use a planning canvas¶
Module: Product and Delivery Teams
Difficulty: Intermediate
Prerequisites: Sessions M7-01 to M7-03, GitHub Copilot app access, GitHub MCP access, and canvas creation access
Duration: 3 hours (1 hour trainer content + 2 hours lab)
Overview¶
Continue the Delivery Decision Studio path. Use the built-in
/create-canvas skill to turn the live Session M7-03 issues into a visible delivery
workspace. Review the generated extension and remove capabilities you do not need.
Then change GitHub state and compare it with the canvas. Detect the stale data
and refresh it.
You describe and operate the canvas without hand-coding the extension.
The GitHub MCP connection supplies issue tools. A plugin may package MCP and canvas capabilities, but you must still review the connection and capabilities. Keep the lab no-code and free of customer or source organization names.
Important
Live canvas creation is required. Stop if /create-canvas, GitHub MCP, or the approved canvas actions are unavailable.
Learning outcomes¶
In this session, you will:
- Decide when a canvas is more useful than another chat response.
- Describe the state and actions needed for a planning canvas.
- Create a canvas from a plain-language request.
- Review and simplify the generated capabilities.
- Load approved GitHub issues into the canvas.
- Detect drift between visible canvas state and a fresh GitHub read.
- Refresh and reconcile the canvas.
- Record ownership, refresh, recovery, and retirement conditions.
- Leave a live canvas linked to the GitHub issues for Session M7-05.
Session structure¶
| Block | Duration | Material |
|---|---|---|
| Trainer content | 1 hour | Trainer guide |
| Lab | 2 hours | Lab guide |
Materials¶
| Resource | Location |
|---|---|
| Slides | slides.md |
| Canvas requirements | lab/starter/ |
Session M3-01 distinction¶
Session M3-01 teaches canvas extension design and implementation. Session M7-04 teaches product and delivery teams to request, review, and operate a canvas through the GitHub Copilot app.