Tringle MCP Studio

Motivation

In our experience, design iteration with AI agents on the major design platforms can be very frustrating. The first few shots are usually very good — then the revisions begin, and it stops being maintainable.

It feels like an endorphin spike that crashes as fast as it arrived: an Alpine ascent straight to the summit, then the long slide back down, with a mountain of rework at the bottom. The friction that used to sit between designers and developers has not disappeared; it has simply morphed into friction between AI agents at different stages.

Tringle is a fundamental rework aimed at turning that cycle of ups and downs into a scenic hike through the Alps — constraint-first wireframes that stay easy to change, with humans and agents working on one canvas through the same propose-and-accept flow.

And one thing we must say: creating designs with AI agents can be a real joy. If you are keen to try it, write to us — and start dreaming.

How the service is run

• A personal project, not a company. The Service is built by a group of like-minded people for their own use — it is not a commercial product, there is nothing to buy, and there is no company behind it.
• Prototype stage. The app is still evolving and changes constantly. Things can break or data can be lost at any time. Do not put anything in it that you cannot afford to lose, and keep your own backups.
• Invite-only. It is a private deployment for authorised users, not an open website.

Terms

• Your work is yours. You own everything you create. The Service only gets the minimal licence it needs to store your projects and show them back to you.
• ‘Public’ means public to other users here — not the internet. Publishing a project lets other account holders view it and clone their own editable copy. Making it private again stops new clones but does not take back copies already made.
• You drive the AI. The Service ships with no AI model. You choose which agent or LLM to connect, via MCP or an API key. That provider’s terms apply, and you are responsible for what your agent does and for reviewing what it proposes. AI output can be wrong.
• No training by us. We do not use your content to train AI models. We cannot speak for the LLM provider you connect — what it does with the data your agent sends and receives is governed by your agreement with that provider, so check the terms you accepted there.
• Be reasonable. No illegal content, no breaking into things, no abusing the API, and no passing off someone’s cloned work as endorsed by them.
• No promises. The Service comes as-is, with no warranty and no liability accepted. Use at your own risk — it is a prototype-stage web app run by like-minded people in their spare time.
• Leaving. Ask us to delete your account and content; the request will be honoured within 30 days. English law governs.

Privacy

• Minimal data. The Service stores your sign-in credentials with your passcode hashed, your session, any API keys you mint hashed as well, and the projects you create. That is what it takes to work — nothing more is collected on purpose.
• Where it lives. The Service operates through a cloud service provider with data stored in Northern Europe.
• No ads and marketing tracking. We run no advertising, no marketing analytics, and no trackers that profile you. The app sets one strictly necessary session cookie; our security and app performance monitoring layer providers may set their own necessary cookies.
• AI providers see what you send them. If you connect an agent, the project data it reads goes to the LLM provider you chose, under that provider’s privacy terms — choose one you trust.
• Security and operations. Like any service, standard operational measures and logs — including errors, performance and access logs, which contains IP addresses — may be kept to run, troubleshoot and protect the Service. They are used for those purposes only, not for advertising or tracking. The hosting layer also keeps its own connection logs under its own policy.

Questions or deletion requests: .