Three ways to connect Figma and Lokalise
You don’t have to choose one approach forever. The Lokalise Figma plugin and the Figma AI agent with Lokalise MCP serve different needs, and both use Lokalise as the translation backbone.
The plugin is structured and fast. It has defined workflows, great documentation, and handles standard localization use cases efficiently. Export frames, create keys, generate language copies, pull translations back — all in a few clicks. If you want a reliable, repeatable process with minimal setup, start here.
The AI agent with MCP is flexible and conversational. You describe what you want in natural language, and the agent coordinates with Lokalise on your behalf. It’s ideal when your workflow doesn’t fit a standard pattern — when you want to sync variables, check for drift across screens, or check on the status of human review tasks directly from within Figma. It’s newer, and it requires a bit more setup, but it lets you define how localization works for your team.
The third connection method is to bring your own LLM. Connect your preferred client to both Figma’s remote MCP server and Lokalise’s MCP server. That way you can orchestrate your translation flows directly in your preferred LLM client without consuming AI credits in Figma.
All approaches connect to the same Lokalise project, the same translation memory, and the same quality infrastructure. They complement each other.
