Use GitHub Copilot Features

Prerequisites

• An active GitHub Copilot subscription (Free, Pro, Pro+, Max, Business, or Enterprise)
• The GitHub Copilot extension installed from your IDE's marketplace
• Authentication with your GitHub account inside the extension
• For Business / Enterprise: organization admin must enable Copilot and assign a seat to the user
• For Free plan: limited access with 50 premium requests/month included

Enable / Disable Granularity (VS Code)

• Click the Copilot status icon in the bottom status bar → Enable or Disable
• When disabling, VS Code asks whether to disable globally or just for the current file's language
• Per-language control also available under Settings → Extensions → GitHub Copilot → Editor: Enable Auto Completions
• You can enable inline suggestions for some languages while disabling them for others (e.g., disable for Markdown, keep enabled for Python)

Ghost Text Behavior

• Suggestions appear automatically after a brief pause while typing
• Triggered by: writing code, writing comments, starting a new function, or naming a variable
• Multiple candidate suggestions exist — you can cycle through them
• If you receive no or limited suggestions, duplication detection may be enabled — this is a common troubleshooting point

Keyboard Shortcuts Reference

Note: Xcode handles full suggestion acceptance via modifier keys rather than Tab. Line-by-line acceptance requires custom shortcut configuration in most editors.

Slash Commands (type / to see full list)

Chat Participants (type @ to invoke)

Chat Variables (type # to invoke)

Image Attachments

• Copilot Chat supports attaching images to prompts: JPEG, PNG, GIF, WEBP
• Useful for describing UI mockups, diagrams, or error screenshots

Model Selection

• Users can switch the AI model Copilot uses for chat responses
• Available models depend on the plan (see Capabilities & Policies page for plan details)
• VS Code also supports model selection for inline (ghost text) suggestions

• Copilot analyses the task and produces a step-by-step plan
• The plan shows which files will be created, edited, or deleted
• You review and approve the plan before any code changes are made
• After approval, you can switch to Agent Mode or Edit Mode to execute
• Best for complex, multi-file work where you want full visibility upfront

• Appear as a sparkle ✨ icon in the editor margin or on selected code
• Provide context-sensitive Copilot actions without opening the chat panel
• Examples: fix this, explain selection, generate docs, review selection
• Available in VS Code and JetBrains IDEs

• Excluded files: Copilot does not offer inline suggestions when editing them
• Excluded files: their content is not sent as context to the model
• Changes to exclusion rules take up to 30 minutes to propagate to IDE clients
• To force a refresh in VS Code: Command Palette → Developer: Reload Window
• To force a refresh in JetBrains / Visual Studio: close and reopen the IDE
• Important caveat: Copilot CLI, Copilot cloud agent, and Agent Mode do not support content exclusion

The Command Palette gives quick access to Copilot functions without memorising shortcuts.

1. Open it in VS Code: Ctrl+Shift+P (Windows/Linux) or Cmd+Shift+P (Mac)
2. Type Copilot to filter available commands
3. Select actions like Explain This, Generate Unit Tests, or Developer: Reload Window

• Log files: Command Palette → Developer: Open Log File or Developer: Open Extensions Logs Folder
• Electron/process-level logs: when nothing useful is in the regular logs, use Help → Toggle Developer Tools
• Network/firewall/proxy issues: open the Command Palette and run GitHub Copilot: Collect Diagnostics — this opens a new editor with connection diagnostics you can inspect or share with support
• Common cause of "no suggestions": content exclusion match, duplication detection, disabled-for-this-language setting, or an authentication/network failure

Install Commands by Platform

Windows additionally requires PowerShell v6+. Requires an active Copilot subscription on all platforms.

Authentication

• On first launch inside a session, run the /login slash command and follow the prompts
• Alternative: a fine-grained Personal Access Token with "Copilot Requests" permission, set via an environment variable
• Token environment variable precedence order: COPILOT_GITHUB_TOKEN → GH_TOKEN → GITHUB_TOKEN
• Exam trap: the "Copilot Requests" permission is only available on user-owned fine-grained PATs — an organization-owned token cannot be used to authenticate the CLI

Interactive Mode

• Start a conversation by running copilot in your project directory
• You're prompted to trust the folder before Copilot can act on it
• Two submodes:
  • Default ask/execute mode — converse and approve actions as you go
  • Plan Mode — toggle with Shift+Tab; Copilot analyzes the request, asks clarifying questions, and builds a structured plan before writing any code
• Auto-compaction: conversation history is automatically compacted once it reaches 95% of the context window

Programmatic Mode

• Pass a single prompt directly: copilot -p "your prompt" (or --prompt)
• Completes the task and exits automatically — no interactive back-and-forth
• Designed for scripts and CI/automation pipelines

• Trusted directories: you explicitly designate which directories Copilot CLI may access and modify
• Tool approval: by default, any command that modifies or executes files requires manual approval at each step
• Approval flags: --allow-tool, --deny-tool, --allow-all-tools
• Caution: --allow-all-tools grants Copilot "the same access as you do to files on your computer" with no per-action approval — use only in trusted, sandboxed contexts
• Supports both local and cloud-based sandbox environments for isolated execution
• Customizable via custom instructions, MCP servers, custom agents, hooks, skills, and persistent memory
• Model configuration: choose GitHub-hosted models, or configure custom providers (OpenAI-compatible, Azure, Anthropic, local Ollama) via environment variables