GitHub Copilot for Developers
Professional developers whose organisation has standardised on GitHub and who want Copilot as an engineering tool rather than an autocomplete
Prerequisites: Working developer setup (terminal, git, an IDE Copilot supports). Module 02 (How to Prompt) or equivalent. A paid GitHub Copilot plan; the cloud agent needs Business or Enterprise with the feature enabled.
Last verified 2026-09-17
Learning objectives
By the end, participants can:
- Name the GitHub Copilot surfaces and plans (Free, Student, Pro ($10 a month), Pro+ ($39 a month), Max ($100 a month), Business ($19 per user a month) and Enterprise ($39 per user a month), each with a monthly allowance of GitHub AI Credits) and pick the surface for a task
- Use the IDE chat modes (Ask, Plan and Agent) and the CLI, and say when agent mode is the right choice
- Delegate an issue to the cloud agent (Copilot cloud agent, which researches a repository, plans, makes changes on a branch in an ephemeral GitHub Actions environment and opens a pull request; on every paid plan, admin-enabled for Business and Enterprise), review its pull request and iterate with @copilot
- Write instruction files (.github/copilot-instructions.md for the whole repository, .github/instructions/NAME.instructions.md for path-specific rules, and AGENTS.md (nearest file wins), with CLAUDE.md and GEMINI.md also read) and a custom agent profile (.github/agents/NAME.agent.md, with name, description, tools and mcp-servers in the front-matter and up to 30,000 characters of instructions below it) that measurably change behaviour
- Connect MCP servers (local and remote MCP servers in the IDEs, Copilot CLI and the GitHub Copilot app, plus repository-level MCP for the cloud agent with the GitHub and Playwright servers enabled by default), configure Copilot code review, and draw the autonomy line for a shared repository
What's in this module
8 lessons · 6 exercises · ~8 min reading
Starts withThe surfaces and the plans~1 minExecutive summary
You will learn how GitHub Copilot's surfaces fit together, from completions and chat through agent mode in the IDE and the CLI to the cloud agent that opens pull requests, and how instruction files, custom agents, MCP servers and code review turn it into a workflow a team can share.
End state is a committed .github/copilot-instructions.md, one custom agent profile, one MCP server wired in, one issue delegated to the cloud agent and merged after review, and a written autonomy line for your repository.
After this module you can choose the surface for a task, delegate scoped work to the cloud agent with confidence, structure instructions so they change behaviour, and keep the review loop honest when the author of a pull request is an agent.
Sources
- GitHub Docs, "Asking GitHub Copilot questions in your IDE", the Ask, Plan and Agent modes and where agent mode runs: https://docs.github.com/en/copilot/how-tos/chat-with-copilot/chat-in-ide
- GitHub Docs, "About Copilot cloud agent", what it does, how it is triggered, and the Actions environment it runs in: https://docs.github.com/en/copilot/concepts/agents/coding-agent/about-coding-agent
- GitHub Docs, "Adding repository custom instructions for GitHub Copilot", the instruction file names and what to write: https://docs.github.com/en/copilot/how-tos/configure-custom-instructions/add-repository-instructions
- GitHub Docs, "Creating custom agents for Copilot cloud agent", the profile path and front-matter: https://docs.github.com/en/copilot/how-tos/use-copilot-agents/coding-agent/create-custom-agents
- GitHub Docs, "About Model Context Protocol (MCP)", MCP across the Copilot surfaces and the cloud agent defaults: https://docs.github.com/en/copilot/concepts/context/mcp
- GitHub Docs, "Getting started with GitHub Copilot CLI", install and first session: https://docs.github.com/en/copilot/how-tos/copilot-cli/cli-getting-started
Prefer the live room?
This module also runs inside our in-person bootcamps and workshops in Copenhagen.
More in Track F
Prompt Engineering Deep Dive
The complete prompting toolkit, from structure and roles to hallucination control and complex industry prompts, plus how to measure prompt quality with evaluations.
Claude Code for Developers
How to run Claude Code as a professional engineering tool: operating manuals, subagents, hooks, skills, MCP, and git/CI workflows.
Agent Harness Engineering
How agents actually work under the hood: the model supplies intelligence, and everything else is a harness you can build.