Learn how GitHub Copilot inline suggestions appear in Visual Studio Code, how to accept full or partial completions, and how next edit suggestions differ from ghost text.
What GitHub Copilot inline suggestions are
GitHub Copilot inline suggestions display proposed code directly in the Visual Studio Code editor. According to the VS Code documentation, they can complete code, comments, tests, and other content while you type.
VS Code documents two forms of inline assistance:
- Ghost text suggestions appear as dimmed text at the current cursor position. A suggestion can complete the current line or propose a larger block.
- Next edit suggestions, also called Copilot NES, predict both the content and location of a possible subsequent edit. The predicted location may be away from the cursor.
The documented prerequisites are an installed copy of Visual Studio Code and access to a GitHub Copilot subscription. GitHub also requires an active Copilot plan, the Copilot extension for the IDE, and a GitHub sign-in within the IDE in its Copilot quickstart.

How to get and accept ghost text
Open a file in a programming language and begin typing. When Copilot produces a completion, it appears as ghost text in the editor. Press Tab to accept the complete suggestion, as described in both the VS Code inline-suggestions guide and GitHub’s IDE quickstart.
To accept only part of a suggestion, use Command+Right Arrow on macOS or Ctrl+Right Arrow on Windows and Linux. VS Code documents this shortcut for accepting the next word or the next line.
Copilot may offer multiple alternatives for the same input. When alternatives are available, hover over the inline suggestion to access them. Their availability is conditional: the documentation says Copilot *might* provide multiple options. GitHub also notes that the exact generated completion can vary.
As an editorial suggestion, review the proposed code before accepting it. This aligns with VS Code’s advice to check proposed APIs, dependencies, and behavior because suggestions can reproduce mistakes as well as useful patterns from surrounding code.

How next edit suggestions work
Next edit suggestions extend inline assistance beyond completion at the cursor. Based on the edits already being made, Copilot NES predicts where another edit may belong and what that edit may contain. A proposal can cover a symbol, one line, or several lines.
To turn the feature on, enable the github.copilot.nextEditSuggestions.enabled setting. When a next edit is available, an arrow in the gutter shows its position relative to the cursor. Hovering over that arrow opens a menu containing shortcuts and configuration options.
The documented VS Code interaction uses Tab twice: press it once to navigate to the proposed edit, then press it again to accept the proposal. This differs from ordinary ghost text, where Tab accepts the completion at the current cursor.
If the VS Code Vim extension is installed, the documentation advises using its latest version to avoid keybinding conflicts with next edit suggestions.
Controls, availability, and plan limits
Inline suggestions can be enabled or disabled from the Copilot menu in the VS Code Status Bar. They can also be controlled for all or selected languages through the github.copilot.enable setting. For a temporary pause, the Copilot menu provides Snooze, which increases the pause in five-minute increments; Cancel Snooze resumes suggestions.
A suggestion can also be requested explicitly by running Trigger Inline Suggestion from the Command Palette. On a connection that VS Code treats as metered, the built-in Copilot provider does not begin new automatic ghost-text or next-edit requests, although requests already underway continue and explicit triggering remains available.
GitHub’s current Copilot plans documentation says inline-suggestion features are available across its listed plans. Copilot Free provides limited access at no cost and is capped at 2,000 completions per month. The same documentation says code completions and next edit suggestions remain unlimited under paid organization and enterprise plans.
Ways to guide and evaluate suggestions
VS Code says Copilot uses the current file and other open files as context. Its documentation recommends descriptive names and focused comments, intended imports, and a small representative example in a related open file. It also advises keeping coding patterns consistent and checking a proposal before accepting it.
Comments can provide hints about the intended implementation, such as an algorithm, design pattern, method, or property. These hints provide context; the documentation does not promise a particular output.
A practical suggestion is to begin with a small, easily checked function and compare each proposal with the behavior you intended. If you keep related files open, use clear names, and state relevant constraints in comments, you are following the contextual practices documented by VS Code. You should still treat every generated result as a proposal requiring review rather than an established correct implementation.
Conclusion
GitHub Copilot inline suggestions in Visual Studio Code comprise ghost text at the cursor and next edit suggestions that can point to another location. Tab accepts complete ghost text, while Command+Right Arrow on macOS or Ctrl+Right Arrow on Windows and Linux accepts a documented partial suggestion. For a next edit, Tab first navigates to the proposal and then accepts it with a second press.
My practical recommendation is to learn ghost-text review and partial acceptance before enabling next edit suggestions. Start in code whose intended behavior you can verify, inspect every proposed API and dependency, and accept only the portion that matches your intent. Enable Copilot NES afterward if location-aware proposals suit your editing workflow.
Sources
- Plans for GitHub Copilot – GitHub Docs
- Inline suggestions from GitHub Copilot in VS Code
- Quickstart for using GitHub Copilot in your IDE – GitHub Docs
Disclosure
This article was prepared with AI assistance.
Improve this guide
Report a problem or suggest the next question
Submissions enter a private editorial queue. They are reviewed and deduplicated before any public update or new page is created.
