Skip to content
GitHub Copilot Train-the-Trainer
Overview

Session M3-01: Copilot app, plugins & canvas extensions

Slides in new tab Download PDF

Module: Agentic Workflows

Difficulty: Advanced

Prerequisites: Sessions M1-01 to M1-03, M2-01 to M2-04

Duration: 3 hours (1 hour trainer content + 2 hours lab)

Overview

Design a canvas extension for one synthetic delivery-readiness board. Start with the contract, then create the canvas in the GitHub Copilot app with the required access. Test direct and agent-requested updates, including a rejected update and recovery. Compare the final canvas with the contract.

A canvas is a shared interactive surface. People use visible controls while an agent uses named capabilities against the same state. The canvas does not replace the system that owns the work.

Learning outcomes

  • Choose a canvas only when visible shared state improves the workflow.
  • Separate canvas state, human actions, and agent-callable capabilities.
  • Write validation rules that reject unsafe or incomplete state changes.
  • Create and review a project-scoped or user-scoped canvas.
  • Recover from a rejected action without weakening the contract.
  • Record ownership, review, failure handling, and retirement decisions.

Deliverables

Submit:

  1. a completed canvas contract;
  2. a canvas created in a sandbox you have approval to use;
  3. checkpoint records for direct, agent-requested, rejected, and recovered changes;
  4. a final state that matches the contract.

Access policy

The lab requires GitHub Copilot app access and the built-in /create-canvas skill. Project scope also requires approval to create files under .github/extensions.

If app access or /create-canvas is unavailable, stop the lab. If project scope is unavailable, use user scope. Do not bypass policy or install a replacement.

Session materials

Resource Purpose
trainer-content/README.md Minute-mapped teaching and demo plan
slides.md Trainer-facing teaching deck
lab/README.md Two-hour learner guide
lab/starter/ Prepared inputs and contract harness
lab/solution/ Worked contract, final state, and completed change records

The default lab remains the delivery-readiness board. An optional Feature Workbench variant uses synthetic CI results to practice a test-gate decision:

See the curriculum glossary for canvas, capability, plugin, and MCP terminology.

Current product references

Product notes were checked against official GitHub documentation on October 5, 2026. Trainers must recheck availability and policy before delivery: