Ghost text

[xhywxtt]

Select Topic Area

Question

Copilot Feature Area

General

Body

I was using Cursor and now switching to Github Copilot. I think Cursor really did good in terms of ghost texting, because its ghost texting is based on the codebase index instead of based on current file context, and do not need your input to do predicting code. How can I get the same experience using Github Copiot? How can ghost text leverage code index resut?

[saddikbannani]

1. How Cursor does it

Cursor uses a codebase-wide index (they run embeddings over the entire project) to let the model "know" about your project context. That’s why its ghost text feels proactive: even without you typing much, it can complete functions or call utilities from other files, because it already indexed them.

2. How GitHub Copilot does it (default)

Copilot primarily looks at the current file context and your recent editor history.

It doesn’t automatically index your entire repo in the same way Cursor does.

Out-of-the-box, ghost text is reactive: it predicts completions based on what you type, not proactively from the whole repo.

3. How to get something closer in GitHub Copilot

Right now, there isn’t a 1:1 equivalent to Cursor’s automatic indexing. But you can extend Copilot’s context in a few ways:

GitHub Copilot Chat with “/workspace” or “/explain” commands:
You can explicitly ask Copilot to reference files/functions across your repo. In VS Code, you can type things like

/explain this function uses utils.js

Copilot Chat then searches your workspace. It’s more manual than Cursor, but closer.

Copilot for Business/Enterprise with “codebase indexing” (Beta):
GitHub is rolling out repo indexing for Copilot Enterprise. This feature makes Copilot aware of the entire repo, similar to Cursor, and lets ghost text + chat draw from the project codebase. If you’re on a personal/free plan, this isn’t available yet.

Manual context feeding:
You can open multiple files side-by-side in VS Code. Copilot then has visibility into those buffers, which improves ghost text quality.

4. What this means for you

If you want Cursor-style proactive ghost text → GitHub Copilot alone won’t give you the same “always-aware” feel unless you’re on Copilot Enterprise with code indexing enabled.

If you’re on the standard Copilot → you’ll mostly rely on inline completions + Copilot Chat, which are contextual but not repo-wide by default.

[anirudhmsu]

Copilot isn't repo wide as compared to cursor (The indexing is much narrower by default).

You could use the copilot's repo level instructions file to provide context about your code base, you can follow this guide: https://docs.github.com/en/copilot/how-tos/configure-custom-instructions/add-repository-instructions.

You could add some context to that like

Project Context

• Database models: src/models/ (Sequelize ORM)
• Authentication: src/auth/ (JWT-based)
• Shared utilities: src/utils/helpers.js
• API routes: src/routes/ (RESTful pattern)

You're manually creating the index of your repo, which would closely match the cursor's index (which is built automatically). Copilot reads this file and uses it to inform suggestions across your entire project. It's not dynamic like Cursor's real time indexing, but it bridges the gap for cross file awareness without needing enterprise features.

If repo wide proactive completions are absolutely necessary to your workflow, standard Copilot won't fully match Cursor's experience yet, but the suggested approach might get you closer.