JDoodle.ai MCP - Build and deploy web apps straight from ChatGPT/Claude
by•
JDoodle.ai MCP connects with ChatGPT/Claude, so you can build websites and web apps directly through chat. You interact with ChatGPT/Claude like normal, while your project is created and updated inside JDoodle.ai with live preview. When you're ready, just ask the AI to publish and JDoodle.ai generates a live link instantly. Unlike one-time generation workflows, you can keep iterating, fixing issues, and shipping updates through chat. Build frontend, backend, or full-stack with database.
Replies
Hey everyone,
ChatGPT and Claude have become a huge part of our workflows.
So we built JDoodle.ai MCP to fit directly into that flow.
JDoodle.ai MCP connects with ChatGPT or Claude or anything that supports MCP, so you can brainstorm, plan, structure, and build websites and apps inside the chat, while seeing the live preview on JDoodle.ai.
As you chat, your project gets created and updated in real time, with instant preview and deployment.
This is not just an initial generation. You can keep asking for changes, refining the UI, adding features, fixing issues, and shipping updates, all directly through ChatGPT or Claude.
Here’s an easy guide on how you can get started: https://www.jdoodle.ai/docs/mcp
Would love your feedback. We’ll be here all day answering any questions.
@gokuljd Congrats! How do you manage version control, collaboration, and rollback for projects created via chat?
@kimberly_ross Thanks Kimberly, we will introduce them in the coming weeks, all the feature on JDoodle.ai will be available through MCP too.
@yerbolat Thanks Yerbolat, we will introduce them in the coming weeks, all the feature on JDoodle.ai will be available through MCP too.
Congrats!
I wonder if we can access to the source code of the projects which we made on the JDoodle
@yaman_alahmad Hi Yaman, yes, we will add an option to download the code.
JDoodle AI MCP is 🔥! Migrated 3 similar no-code→fullstack (Bubble→React+Supabase). Backend scaling hurt most. Tips? Paweł @ Inigra (Clutch 5★)
@pawel_inigra Thanks Pawel. Is it a code management or cost with scaling? Happy to suggest on your specific situation. Feel free to ping me directly.
This is interesting. How does the MCP server handle the deployment piece? I've been playing with MCP for code generation but deployment's always the tricky part. What's the flow like from Claude to live app?
@mykola_kondratiuk Thanks Mykolo, deployment is handled by JDoodle.ai like it does for the apps created directly on our platform. MCP just triggers that endpoint.
chat-driven codegen plus “publish a live link” will quickly run into state drift and unsafe execution, especially when multiple iterations touch backend, DB migrations, and secrets in the same session. Best practice: treat each chat change as a Git commit and run it in an ephemeral sandbox (devcontainers or microVMs like Firecracker or gVisor) with CI checks before deploy, and follow MCP OAuth scoping and tool allowlists for least privilege. Question: when MCP requests modify code and schema, do you generate deterministic migration plans with rollback and a diff or PR style review, or is state tracked only inside JDoodle projects?
@ryan_thill Thanks Ryan, This is what we do for the projects created on JDoodle.ai, runs on a dedicated sandbox, every change is a git commit, database changes have forward and rollback for each change, test before deploy/commit, etc. MCP projects will follow similar pattern eventually.
@gokuljd That’s a solid foundation. Dedicated sandbox plus “every change is a git commit” and forward/rollback DB migrations with tests before deploy is exactly how you avoid drift as chat iterations pile up. One best-practice add for MCP is policy-based tool allowlists and secret scoping per environment so model-driven changes can’t accidentally widen blast radius. Open question: will MCP expose a PR-style diff/preview and staged deploys (dev to prod) so teams can review and roll back deterministically?
Do you have any demos of things built purely with JDoodle.ai? It would be great to see the initial prompts and the chat history.
Building and deploying directly from a chat interface removes so much friction from the development process. Curious about the deployment infrastructure — where do the apps get hosted, and how does scaling work?
@gokuljd Congrats on the launch. Closing the loop between chat and live deployment is a powerful shift, especially if projects can keep evolving instead of being one-shot generations.
I’m curious how you handle trust boundaries when the model can modify backend logic and database schema through MCP. If a prompt or imported snippet contains adversarial instructions, what safeguards prevent unsafe migrations or unintended code paths from making it into a live deployment?