Hi, I'm Mona 🐙
Today, we're going to build a simplified version of the Copilot Playbook site open in the other tab, from scratch, together with Copilot.
In this workshop, you'll get hands-on with the simple concepts introduced in the Copilot Playbook. Following the PLAN → CODE → REVIEW → TEST & SECURE → OPERATE phases, you'll use GitHub Copilot to ultimately build a simplified version of the Playbook together with Copilot.
1.1 — Today's goals
- PLAN: Equip Copilot with "context", "rules", and "shape" via MCP, Instructions, and Skills
- CODE: Design and implement the site with Copilot Chat (Plan / Agent), and add extra features with Copilot Agent
- REVIEW: Automatically review PRs with Copilot Code Review
- TEST & SECURE: Guarantee quality and safety with Dependabot, CodeQL (GHAS), and GitHub Actions (Playwright)
- OPERATE: Build AI into daily operations with Agentic Workflow
1.2 — For administrators
This section is for Enterprise administrators. Below is a summary of the Enterprise / Organization settings to enable in advance, so participants can complete the entire hands-on without issues.
Reminder: These settings must be applied to the Enterprise / Organization that participants get their Copilot license from. Enabling them on a different Enterprise / Organization will not take effect for the participants.
【Required】
- One test Organization is required. Participants will use it to freely create test repositories for the hands-on.
【Features to enable at the Enterprise level (best effort)】
※ If the items below are not allowed at the Enterprise level, they cannot be enabled at the Org level.
-
Codespaces
- Path: Your Enterprise → Policies → Codespaces → Enable for all organizations
- This is so participants use the same environment as I do. Local PC is possible as a fallback, but Codespaces is recommended.
-
Actions
- Path: Your Enterprise → Policies → Actions → Enable for all organizations → "Allow all actions and reusable workflows"
-
Copilot Code Review
- Path: Your Enterprise → AI Controls → Agents → Available Agents → Copilot Code Review → Enabled
-
Copilot Cloud Agent
- Path: Your Enterprise → AI Controls → Agents → Available Agents → Copilot Cloud Agent → Enabled
-
Copilot CLI / Chat / Agent Mode
- Path: Your Enterprise → AI Controls → Copilot
- Copilot CLI → Enable everywhere
- Copilot Agent Mode in IDE Chat → Enable everywhere
- Auto-Approval in Copilot Agent Mode → Enable everywhere
- Path: Your Enterprise → AI Controls → Copilot
-
Partner Agents
- Path: Your Enterprise → AI Controls → Agents → Partner Agents (Preview)
- Anthropic Claude → Allowed
- OpenAI Codex → Allowed
- Path: Your Enterprise → AI Controls → Agents → Partner Agents (Preview)
-
MCP
- Path: Your Enterprise → AI Controls → MCP
- MCP servers in Copilot → Enable everywhere
- Path: Your Enterprise → AI Controls → MCP
【Features to enable in the test Organization (best effort)】
-
GitHub Codespaces
- Path: Org Settings → Codespaces → General → User permissions → "All members"
- This is so participants use the same environment as I do. Local PC is possible as a fallback, but Codespaces is recommended.
-
GitHub Actions
- Path: Org Settings → Actions → General → "Allow all actions and reusable workflows"
- Required for Workflow execution and for the Copilot Code Review / Cloud Agent / Agentic Workflow features below.
-
Copilot Code Review
- Path: Org Settings → Copilot → Policies → Features → "Copilot code review" → Enabled
-
Copilot Cloud Agent
- Path: Org Settings → Copilot → Policies → Features → "Copilot cloud agent" → Enabled
-
Copilot CLI
- Path: Org Settings → Copilot → Policies → Features → "Copilot CLI" → Enabled
-
Permission to create repositories
- Path: Org Settings → Member privileges → Repository creation → ☑ Private (Members will be able to create private repositories)
【Skippable】
- The security-related part (GHAS) can be skipped with no impact on the rest of the hands-on.