Coming from Cursor? Keep Cursor.

Keep your editor: install Celiador Remote

1. In Cursor, search for Celiador Remote in the Extensions view and install it. In VS Code, download the VSIX from the latest release and run Extensions: Install from VSIX…. It is also on Open VSX.
2. Sign in from the Celiador item in the status bar (email, magic link or GitHub).
3. Follow the start guide from step 2: connect GitHub, put the queue on your phone, open the repository, label an issue.

Runs happen on your machine in a git worktree, the same as in the Celiador editor. The extension uses the Claude Code CLI you already have, or downloads it once. Your agent usage goes through celiador.ai on our key with your account’s limits; an Anthropic key in your environment is not used.

Or move to the Celiador editor

1. Download Celiador for macOS or Linux and open it.
2. Run Celiador: Import Settings from Another Editor from the command palette. It finds Cursor or VS Code on your machine and brings your setup over.
3. Sign in, open a folder, and label an issue.

What comes with you

• Settings: merged into yours. Anything you already set in Celiador wins.
• Keybindings and snippets: added, never overwriting what you have.
• Extensions: installed from Open VSX, the open extension registry. The import tells you by name which ones it could not find there.

What deliberately doesn’t

• Cursor’s own settings, Cursor-only extensions and Copilot. The Celiador editor has its own agent built in; two AI layers in one editor fight each other.
• Your Cursor account, chat history and indexes. Those stay with Cursor. Celiador doesn’t build a server-side index of your code at all.

What’s different here

• The unit of work is an issue, not a chat. You brief the agent the way you brief a colleague, it works on a branch of its own, and the result is a pull request your team can read. The run narrates itself on the issue.
• The review happens on your phone. What needs you, what is running, what is ready: allow or deny a risky command, answer a question, send a follow-up, approve and merge. The editor’s Agents view shows the same.
• Claude is the agent, by design. Not one option in a model menu that can change under you.
• One pricing truth. The limits we enforce are published as numbers, live from the same service that meters you: see pricing.
• We never train on your code, and your code stays local: privacy, in plain words.

If you’re leaving because of billing

Most developers who switch tools don’t leave over features. They leave when a plan changes under them, a limit appears that wasn’t there before, or a model they relied on disappears from the menu. We can’t promise we’ll never change a price. We do promise how we’ll do it: dated notice, grandfathering with an end date, and cancelling as easily as you joined. That’s written down in our conduct pledge, and you can hold us to it.