What's changed

• you now upload files instead of submitting URLs to them
• you can process multi-page PDFs as well as images
• you can browse and delete previous forms that you've processed
• you can open the generated form in a new tab

Full Changelog: v0.1-alpha.0...v0.2-alpha

Features:

• User submits URL of a JPEG image of a form
• Image is sent to an LLM, along with a prompt and JSON schema
• Uses Anthropic LLM by default, or OpenAPI if no Anthropic API key is found
• LLM returns JSON representing any form elements in the image
• JSON is used to generate a multi-page web form in an iframe

Thanks to joelanman, Jarek-Rolski, eltociear and simonw for contributions!