Session 09 — Copilot Cloud Agent¶
Module: Agentic Workflows Difficulty: Advanced Prerequisites: Sessions 01–07 Duration: 3 hours (1 hr trainer content + 2 hr lab)
Overview¶
This session shows how to use the GitHub Copilot cloud agent when policy and availability permit it. It covers issue preparation, human review, session management, and a shared protocol for comparing approved third-party agents.
Learning Outcomes¶
- Prepare a repository for the Copilot cloud agent
- Write issues the cloud agent can complete
- Compare approved third-party agents using a shared evaluation rubric
- Manage agent sessions and track progress
- Review agent-generated PRs and request changes
Session Materials¶
| Resource | Location |
|---|---|
| Lab exercises | lab/ |
| Lab starter files | lab/starter/ |
| Lab solution files | lab/solution/ |
Key Topics¶
- Cloud agent: issue-driven and asynchronous
- Configuration:
copilot-setup-steps.yml, firewall, runners - Writing effective agent issues
- Third-party agent comparison: access, policy, quality, safety, reviewability, and metering
- Agent session management
- Security: network firewall, toolset restrictions