Track: Product and Delivery Teams

Target Audience: Product Managers, Product Owners, Business Analysts, Project Managers, delivery leads, and other non-developer roles Duration: 18 hours (6 sessions × 3 hrs) Difficulty: Beginner → Intermediate Recommended Delivery: 3 days

---

Description

Use the GitHub Copilot app to work through one product initiative across six sessions. The path is called Delivery Decision Studio. That is the curriculum's name, not a GitHub product or app mode.

Gather context and settle product decisions first. Then plan the issues, build a planning canvas, supervise parallel reviews, and draft a follow-up within the agreed limits.

The initiative is the fictional Service Request Portal. Do not replace it with customer work in the standard path. Trainer-approved substitutions must be synthetic, sanitized, and free of customer or source organization names.

Prerequisites

• A GitHub account with GitHub Copilot access
• Browser access to the training repository
• The GitHub Copilot app
• The approved GitHub MCP server with access to the training repository
• Canvas creation, orchestration, and Automations access
• Basic knowledge of the learner's product or delivery process

You do not need programming experience, Git command-line knowledge, or repository administration skills.

GitHub Copilot access is required for Sessions M7-01 to M7-06. Learners without access should not take this track. Stop and reschedule their training after access is available.

Capability setup

Complete this setup before Session M7-01. Check the tools again before each later session. Organizational policy can change which tools the app makes available.

1. Open the training project

1. Sign in to the GitHub Copilot app.
2. Add or open the approved training repository under Projects.
3. Start an interactive session.
4. Ask Copilot to read one known file from the repository.
5. Stop if it cannot read the approved project.

2. Inspect installed customizations

Open Customize, then Installed. Record the available:

• skills;
• MCP servers;
• plugins;
• canvases.

Use only capabilities approved for the training repository.

An MCP server and a plugin do different jobs:

• MCP server: exposes named tools and data sources to Copilot.
• Plugin: installs a reusable package that may include skills, agents, hooks, MCP servers, or canvases.

A plugin installation does not tell you whether GitHub MCP can read the training repository. Test the connection separately.

3. Check skills

Sessions M7-04 and M7-05 use the built-in /create-canvas and /orchestrate skills. Type / in a session and check that both commands appear before those sessions. They are built-in app skills, so do not install look-alike community skills.

Session M7-02 supplies a project skill under .github/skills/decision-interview/SKILL.md. After copying it into the training repository:

1. commit the skill;
2. open a new app session for the project;
3. open Customize → Skills → Installed;
4. check that decision-interview appears;
5. run /decision-interview with a short test request.

Copilot CLI users can check the same project skill with:

/skills reload
/skills info decision-interview

4. Check GitHub MCP

Sessions M7-03 to M7-06 require GitHub repository tools.

In the app:

1. open Customize → MCP → Installed;
2. check that the approved GitHub MCP connection is available;
3. start a session and ask Copilot to list the available GitHub tools;
4. test one read against the training repository;
5. do not continue to a write lab until the read succeeds.

GitHub MCP is built into Copilot CLI. CLI users can inspect it with:

/mcp list
/mcp show github

If an additional MCP server is required, install it through Customize → MCP or /mcp add using the organization-approved server definition. Do not paste personal access tokens into prompts or repository files.

5. Check session-specific tools

Session	Required tool or access
M7-01	App project and repository reads
M7-02	Project skill decision-interview
M7-03	GitHub MCP issue reads and writes
M7-04	Built-in /create-canvas and GitHub MCP reads
M7-05	Built-in /orchestrate, child-session steering, and GitHub MCP
M7-06	Automations and GitHub MCP

Stop the lab if a required tool is unavailable. Role-play and prepared files do not replace hands-on work.

6. Check session controls and context syntax

Before Session M7-01, check these controls in the current app:

• session modes: Interactive, Plan, and Autopilot;
• @ for adding files, # for referencing issues, and / for commands;
• /context for current context usage;
• /compact for summarizing an active session under token pressure;
• /clear or /reset for a fresh session;
• /rubber-duck for an independent critique from a different model.

GitHub notes that command availability depends on context. Type / in the app before delivery. If a command is missing, record it in the trainer notes. Do not teach a substitute without checking that it works.

Delivery Decision Studio continuity

Use these roles throughout the path:

Role	Responsibility
Decision owner	Makes product and scope decisions
Delivery reviewer	Checks results, readiness, and handoffs
Work-system steward	Checks the authoritative GitHub or Azure Boards state
Trainer	Checks access and teaches optional practices

Save and reuse these results:

Session	Authoritative result	Used next
M7-01	initiative-context-brief.md	Repository context for Session M7-02
M7-02	decision-brief.md, issue-proposal.md	Product decisions for Session M7-03
M7-03	Live parent and child GitHub issues	Issue state for Session M7-04
M7-04	Live canvas linked to GitHub issues	Reconciled planning view for Session M7-05
M7-05	orchestration-plan.md, live issues, reviewed parent result	Authoritative inputs for Session M7-06
M7-06	automation-definition.md, stakeholder-update.md	Human-owned operating decision

At each step, record the next action and open questions in the file or work system listed above.

What learners should bring

• The supplied Service Request Portal initiative
• The supplied synthetic issue and status data
• The approved live issue state from the training repository
• One recurring workflow pattern they want to evaluate
• Their team's definition of ready or done
• One decision that must remain with a named person

The supplied scenario covers every exercise and is the default learner path.

Sessions

Order	Session ID	Title	Difficulty	Duration
1	M7-01	Work Effectively in the GitHub Copilot App	Beginner	3 hrs
2	M7-02	Interview Ideas and Shape Better Work	Beginner	3 hrs
3	M7-03	Plan Work with GitHub Issues and MCP	Intermediate	3 hrs
4	M7-04	Build and Use a Planning Canvas	Intermediate	3 hrs
5	M7-05	Orchestrate Agents and Workstreams	Intermediate	3 hrs
6	M7-06	Automate Delivery Follow-up and Connect Work Systems	Intermediate	3 hrs

Recommended delivery schedule

Three-day workshop

Day	Sessions	Focus
Day 1	Sessions M7-01 to M7-02	App fluency, context, and interviewing
Day 2	Sessions M7-03 to M7-04	GitHub issues, MCP, and planning canvases
Day 3	Sessions M7-05 to M7-06	Orchestration, automation, and work-system handoff

Learning outcomes

After this track, you should be able to:

1. Compare unscoped and repository-grounded Copilot work.
2. Install and use a project skill to create and revise a decision brief.
3. Use GitHub MCP to preview, create, check, and correct an issue plan.
4. Create a planning canvas and reconcile it after GitHub state changes.
5. Use built-in orchestration to supervise, redirect, and review child sessions.
6. Run a draft-only Automation twice and compare the recorded results.
7. Decide which work system approved by the organization remains authoritative.

Trainer guidance

• Use the supplied Service Request Portal path in standard delivery.
• Keep customer and source organization names out of prompts, artifacts, issues, screenshots, and demonstrations.
• Reuse the initiative so learners can follow the work between sessions.
• Check current official GitHub documentation and organizational policy before every live delivery.
• Treat generated output as a proposal. The accountable person accepts, rejects, or revises it.
• Keep the path no-code. Learners review generated extensions and tool calls but do not write application code, MCP servers, plugins, or canvas extensions.

Trainer-only Azure Boards supplement

Use Azure Boards only when the delivery plan calls for it before the workshop. Do not introduce it as an alternate learner path during the standard labs.

The tools described in scripts/azure-boards/README.md seed, check, and reset a synthetic backlog for Agile or Scrum projects. They check existing project settings and never create them.

Use the Session M7-02 supplement to shape a seeded requirement. Use the Session M7-05 supplement to supervise seeded workstreams. Both still require GitHub Copilot.

Next steps

• Agentic Development teaches technical teams how to build the agents, skills, and integrations used behind these workflows.
• Full Mastery prepares trainers to deliver every technical and role-based session.