Skip to content

Claude Code

Frontend lets you run Claude Code directly from your project’s built-in terminal. Claude operates inside a sandboxed environment with full access to your project’s codebase, so you can use it to build, refactor, and debug just as you would locally — without any local setup.

To use Claude Code, open the terminal in your project and launch Claude. You’ll be prompted to authenticate with your existing Claude account — no new account required. Once logged in, Claude has access to any Claude.ai connectors or MCP servers you’ve already configured, so your existing workflows carry over automatically.

Once authenticated, Claude runs in the sandbox with direct access to your project files and can:

  • Read and edit any file in your codebase
  • Run terminal commands and scripts
  • Use your connected MCP servers and Claude.ai connectors
  • Work with Git to stage, commit, and manage changes

Claude has access to Git inside the sandboxed environment, but the storage is ephemeral — local changes do not persist between sessions. To keep your work:

  • Push to a remote repository — commit and push any changes to GitHub or another remote before your session ends
  • Create a version snapshot — use the createVersion tool available in the Frontend MCP server provided to Claude; this saves a named snapshot of your project that you can restore at any time from within Frontend

Making it a habit to push or snapshot before closing a session ensures nothing is lost between work sessions.