GitHub Copilot — AI Coding Assistant
Complete guide to GitHub Copilot, the AI coding assistant. Covers prompt engineering, configuration, copilot-instructions.md, agent mode, and GitHub workflow integration.
GitHub Copilot — AI Coding Assistant
GitHub Copilot is the most widely adopted AI coding assistant, used by millions of developers across VS Code, Visual Studio, JetBrains IDEs, and now the terminal via Copilot CLI. It's built on OpenAI models but distinguished by its deep integration with the GitHub platform — it understands your repos, PRs, issues, and Actions, not just the file you're editing.
Where Cursor and Windsurf replace your editor, Copilot layers onto the editors and workflows you already use. That makes it the lowest-friction choice for teams already living inside GitHub.
How Copilot Works
Copilot combines several interaction models into one tool:
- Inline completions — As you type, Copilot suggests the next lines. The best-known feature, and still the fastest way to get value.
- Copilot Chat — An IDE chat panel that answers questions about your code, with the ability to generate code, explain snippets, and suggest fixes.
- Agent mode — An autonomous mode that plans and executes multi-file changes with terminal access, added in 2025.
- Copilot CLI — A terminal-based agent for prompts, piped I/O, and scripting workflows outside the IDE.
- GitHub context — When connected to GitHub, Copilot can reference open issues, pull requests, and repository structure in its answers.
Because Copilot is managed by GitHub/Microsoft, it also ships with organization controls: policy management, code referencing settings, and audit logs — which is why it's a common default in larger companies.
Core Features
- Inline completions — Context-aware code suggestions in real time, powered by your open files and project.
- Copilot Chat — Ask questions, refactor code, generate tests, and get explanations without leaving your editor.
- Agent mode — Autonomous multi-file editing with terminal access, keeping you in the review loop.
- Copilot CLI — Terminal-first agent for scripting and CI-friendly workflows.
- GitHub integration — PR/issue context, code review suggestions, and Actions support.
- copilot-instructions.md — Project-level custom instructions that shape every Copilot response, similar to
.cursorrulesorCLAUDE.md. - Multi-IDE — VS Code, Visual Studio, JetBrains (IntelliJ, PyCharm, WebStorm, etc.), Neovim, and CLI.
Quick Start
Install the GitHub Copilot extension in VS Code or your JetBrains IDE, then sign in with your GitHub account.
# Copilot CLI (optional, for terminal workflows)
npm install -g @github/copilot-cli
copilot "explain this repo architecture"
Then:
- Open a project and press
Tabto accept your first completion. - Add a
copilot-instructions.mdfile in your repo root with your team's conventions. - Open Copilot Chat (
Cmd+Shift+I/Ctrl+Shift+I) and ask a question about the codebase. - For bigger tasks, switch to Agent mode and watch it plan a multi-file change.
Writing Better Prompts
Copilot performs noticeably better when you give it structure:
- Be specific about the target — "Write a function that parses ISO-8601 durations" beats "write a parser."
- Provide examples — Show one input/output pair in your prompt or in a comment above the code.
- Reference your own code — "Use the same error handling as
utils/http.ts" produces consistent code. - Keep
copilot-instructions.mdcurrent — the file you point at architecture, naming, and testing conventions shapes every response.
For a deeper workflow, see Prompt Engineering in Copilot. For settings and copilot-instructions.md, see Configuration.
Copilot vs Cursor vs OpenCode
| Feature | GitHub Copilot | Cursor | OpenCode |
|---|---|---|---|
| Form factor | IDE extension + CLI | IDE (VS Code fork) | Terminal agent |
| Tab completion | ✓ | ✓ | ✗ |
| Chat in editor | ✓ | ✓ | ✗ |
| Autonomous agent | ✓ | ✓ | ✓ |
| GitHub integration | Deep (PRs, issues, Actions) | Partial | Partial |
| Custom rules | copilot-instructions.md | .cursorrules | AGENTS.md |
| Open source | ✗ | ✗ | ✓ |
| Local models | ✗ | Via API keys | ✓ (Ollama etc.) |
If you live in GitHub, Copilot is the most integrated option. If you want a full agent-first editor, Cursor. If you want a free, open-source, terminal-based agent, OpenCode.
Pricing
- Free — Limited completions and chat for individuals.
- Pro — $10/month for unlimited completions and higher chat limits.
- Business — ~$19/user/month with org policy controls, audit logs, and code referencing safeguards.
- Enterprise — Custom plans with SSO, and deployment options.
Frequently Asked Questions
Related Resources
- Prompt Engineering in Copilot — get better results from Chat and Agent mode
- Configuration — master
copilot-instructions.mdand workspace settings - Tool Comparison — Copilot vs Cursor vs OpenCode vs Claude Code side-by-side
- Claude Code — Anthropic's terminal-based agent, an alternative for prompt-first workflows
Related Articles
Antigravity CLI — AI Coding Agent for Everyone
Google's terminal-based AI coding agent. Replaces Gemini CLI for individual consumers. Go-native, multi-model, async workflows. Install with curl, authenticate with Google, start coding.
CLI-First Development with Antigravity Image Pipeline
Use fast CLI tools for all coding, then switch to Antigravity for Nano Banana image generation. Both under one Google AI Pro plan with no per-image fees.
Jcode — Getting Started
Install and configure jcode, the RAM-efficient open-source AI coding agent in Rust. Covers install, provider login, memory, swarms, and your first session.