I’m exploring ways to build interfaces agentically by integrating WeWeb into an AI-driven workflow. I noticed that the v2/designs/{id}/source_codes/coded_components endpoint is being used by the internal WeWeb AI to generate coded components, which is exactly the kind of access I’m hoping to experiment with.
From what I can tell, this endpoint—and others like it—are available behind the scenes but not yet part of the public API. Is there any way to request early (alpha/beta) access to these design-related endpoints so that external agents can orchestrate component creation directly?
I saw updates suggesting this might be part of an upcoming release (the AI chatbot reported the ETA as January), but I’d love to test it out ahead of time if possible. If there’s someone specific I should reach out to (like @joyce ?), happy to tag them here as well.
Appreciate all the progress your team is making! Thanks!
Thanks @Agustin_Carozo - For some broader context around this, seems like platforms (Xano, FlutterFlow, for a few examples in the no code space) are publishing their own MCP servers to give client agents (chatgpt, claude, etc… or even coding clients like windsurf, cursor, etc…) the ability to view and interact with tools (ie: create a new component or clone this existing interface in WeWeb). MCP just functions as an AI agent standard for interacting with these tools that are likely closely tied to individual api endpoints.
So… it would be great if WeWeb could either…
Make the api public to view/create/edit comonents and pages so interested developers could use an agent to make changes on their behalf via the api
Or create an MCP server to expose those “tools” (potentially without exposing the api the powers it) that would be immensely valuable and speed development considerably!
Right now, we’re forced to go through the WeWeb AI chat experience to leverage any of the behind-the-curtain magic and it’s likely that (with some documentation) a public api or mcp tools could be incredibly useful when used with these clients that have impressive models capable of great orchestration!
“We’re experimenting with an official FlutterFlow MCP Server that allows you to read and write to your projects, create new pages and components, accomplish refactors, and more, right from your favorite AI tools!”
and it’s likely that (with some documentation) a public api or mcp tools could be incredibly useful when used with these clients that have impressive models capable of great orchestration!