What is Layntra?
Layntra is an open-source Codex plugin and local Figma companion. It lets you inspect a Figma target, review a plan, explicitly apply supported editable changes, and undo the last guarded write.
Straight answers
Layntra FAQ.
What it is, what it costs, what stays local, what Figma still controls, and what a nontechnical user needs to install.
Is Layntra free?
Yes. Layntra is free and open source under the MIT License. Version 0.1.0 has no hosted Layntra account or paid Layntra plan.
Does Layntra bypass Figma plan or seat restrictions?
No. Layntra does not bypass Figma permissions, organization policies, seat types, or plan restrictions. Its companion must run where Figma Design plugins are permitted.
Does Layntra require a Figma API token?
No. Layntra uses a Figma companion plugin and a bridge bound to local loopback, so it does not ask for a Figma API token.
Is Layntra fully offline?
No. The Layntra bridge between Codex and Figma is local, but Codex model access and data handling follow your Codex configuration.
Can a product manager use it without writing code?
After installation, the normal workflow uses explicit natural-language commands in Codex. Initial setup requires cloning the repository, running one install script, and importing a manifest file in Figma Desktop. The illustrated guide shows the exact path.
Are the results editable in Figma?
Yes. Supported output is created as editable Figma nodes rather than a flattened screenshot.
Why does Layntra use $layntra?
The explicit command makes activation, target, mode, and write intent visible. Regular conversation stays regular conversation; it does not silently become permission to modify Figma.
What can version 0.1.0 change?
Version 0.1.0 supports bounded creation and updates for common editable nodes, with batches limited to 100. It does not expose deletion or arbitrary code execution.
What if the page or selection changes after planning?
Layntra refuses the write. Run status or plan again against the new target instead of letting the system guess.
Where do I report a problem?
Open a public GitHub issue with the command you ran, what you expected, and any safe-to-share error text.