Started in Lovable, continuing in Claude Code? Read this first

Almost everyone who builds with AI ends up using more than one tool: start in Lovable or Bolt for speed, move to Cursor or Claude Code for control. The move itself is where projects get hurt, because each tool assumes it owns your project.

The trap

Your project's history lives inside whichever tool made it: Lovable's version list, Bolt's chat, Cursor's checkpoints. None of them can see the others. Switch tools naively and you start a second, disconnected life for the same project: two histories, no way back across the boundary, and the old tool's hosting still serving the old version.

The safe move, step by step

1. Get the project out as real files first. From Lovable or Bolt, turn on GitHub sync (both support it, in project settings). This is the bridge: your project becomes ordinary files any tool can edit.
2. Pick where the project lives: one place both tools answer to. That can be raw GitHub if you're comfortable there, or Plyo, which imports via that same sync and keeps a plain-English timeline of every version regardless of which tool made it.
3. Point the new tool at the home, not at a copy. For Claude Code or Cursor, that's one paste (exact setup per tool). Now both tools write to the same history.
4. Move the live site last. Once the new workflow feels right, publish from the home and retire the old tool's hosting, so there's exactly one current version in the world.

The one rule

Copies rot; homes don't. Downloading a zip and re-uploading it into the next tool works exactly once, then leaves two diverging versions of your project on your hands. Sync into a home once, and every future tool switch is a connection, not a migration.