GitHub Copilot

GitHub Copilot is GitHub's AI coding assistant for software developers. GitHub's documentation describes it as "an AI assistant that helps you write, understand, and ship software" 2. It suggests code as you type, answers questions about a codebase in chat, edits files across a project in agent mode, and can take an assigned issue and come back with a pull request 1 4.

Where Copilot runs

Copilot works inside the tools developers already use. GitHub lists extensions for Visual Studio Code, Visual Studio, JetBrains IDEs, Neovim, Vim, Eclipse, Azure Data Studio and Raycast, along with the terminal, GitHub.com and GitHub Mobile 1. Which features you see depends on your plan, the client and your organisation's policies 2.

Main features

• Code suggestions: inline completions, from a single line to whole functions, as you type 2
• Copilot Chat: ask about the code in front of you, get explanations, generate tests or fixes 1 2
• Agent mode: Copilot works through a multi-step change and edits several files in your workspace, with approval controls along the way 1
• Cloud agent: assign an issue to Copilot and it works on a branch in its own environment, then opens a pull request for review 4
• Code review: Copilot reviews pull requests and proposes edits 1 2
• Model choice: paid plans can pick between models from providers such as OpenAI, Anthropic and Google 1 3

The cloud agent in more detail

When you assign an issue to Copilot, it researches the repository, writes a plan and makes changes on a branch inside an ephemeral environment powered by GitHub Actions, where it can run the project's tests. It then opens a pull request, and a person reviews and merges it 4. Typical tasks are bug fixes, small features, extra test coverage, documentation updates and merge conflicts. The agent is available on all paid plans, but on Business and Enterprise an administrator has to enable it first 4.

The agent can use Model Context Protocol servers to reach other tools and data. The GitHub MCP server and the Playwright MCP server are enabled by default 4.

Plans and pricing

GitHub offers a free tier with limited features and 2,000 inline completions a month, plus individual Pro ($10 a month), Pro+ ($39 a month) and Max ($100 a month) plans 3. According to GitHub's Copilot plans documentation, organisations buy Copilot Business ($19 per seat a month) or Copilot Enterprise ($39 per seat a month) 3. Verified students get Copilot at no cost, and teachers and some open-source maintainers may qualify for free Pro 3. Paid plans include a monthly allowance of usage credits for premium models and agent work, so check the plans page for current limits 3.

What the research says

GitHub's best-known productivity study was a controlled experiment with 95 professional developers who built an HTTP server in JavaScript. The group using Copilot finished 55% faster on average (1 hour 11 minutes against 2 hours 41 minutes), and 78% of them completed the task compared with 70% without it 5. That result covers one coding task by developers. It is not a measure of marketing or business productivity.

Copilot and other AI coding tools

Copilot sits alongside editor-based assistants such as Cursor and Windsurf, and terminal agents such as Claude Code and OpenAI Codex CLI. GitHub also lets you assign tasks to third-party agents such as Claude and OpenAI Codex from the same interface 1. Many of these tools read shared instruction files like AGENTS.md, so a team can keep one set of project rules.

Using Copilot with a publishing API

Developers who automate social media often use Copilot to write the scripts that call a publishing API. A developer can also connect a social publishing server to Copilot in VS Code. Publora, a social media scheduler and publishing API, documents MCP setup for VS Code and other clients, with 18 tools covering 10 platforms 6.