Noureddine RAMDI / Vibe Workflow: Embeddable node-based AI workflow builder for generative image and video pipelines

Created Sat, 23 May 2026 20:41:14 +0000 Modified Sat, 23 May 2026 20:41:27 +0000
959 Words
github-stars javascript nextjs fastapi node-based-ui generative-ai workflow-builder

SamurAIGPT/Vibe-Workflow

Vibe Workflow tackles a problem many developers face when building AI-powered creative tools: how to visually compose complex generative image and video pipelines in a flexible, reusable way. Instead of relying on proprietary platforms or monolithic SaaS solutions, Vibe Workflow provides an open-source, MIT-licensed node-based editor that you can self-host or embed into your own React or Next.js applications.

what vibe workflow does and its architecture

At its core, Vibe Workflow is a node-based AI workflow builder designed to visually compose generative pipelines for images and videos. It positions itself as a direct open-source alternative to platforms like Weavy AI, Krea Nodes, Freepik Spaces, and FloraFauna AI, all of which offer similar visual node editors but are proprietary.

The project is structured as a monorepo with three main parts:

  • client: A Next.js frontend application that provides the web UI.
  • packages/workflow-builder: A standalone, shared UI library that implements the core node editor and workflow builder. This is the key piece you can extract and embed into any React or Next.js app.
  • server: A FastAPI backend that handles workflow execution and connects to MuAPI (Vadoo AI) for the generative AI capabilities.

This separation means the node editor is decoupled from the backend and the default frontend app, enabling you to reuse the visual canvas independently. The backend requires a MuAPI API key to access the generative AI models powering image and video generation.

The stack uses:

  • Next.js for the frontend, benefiting from React’s ecosystem and server-side rendering.
  • FastAPI (Python) for the backend API, providing a performant and modern async web server.
  • Node.js/npm workspaces for managing the frontend and shared UI library dependencies.

The project supports running locally with Node.js and Python or via Docker Compose for simpler deployment.

technical strengths and design tradeoffs

What sets Vibe Workflow apart is its modular, monorepo design and the standalone workflow-builder UI library. This library encapsulates the entire node-based editor, allowing developers to embed a full-featured generative AI pipeline editor directly into their own SaaS products or web apps without dragging in the backend or the default frontend.

The monorepo setup using npm workspaces links the client and the UI library cleanly, which is a solid developer experience (DX) choice. It allows simultaneous development and sharing of components without duplication.

The backend separation via a FastAPI service means the AI model provider (MuAPI) is abstracted away from the UI, making it easier to swap or update the AI service integration if needed.

One tradeoff is the dependency on MuAPI (Vadoo AI) for generative models. While this centralizes AI capabilities, it ties you to an external API key and service. However, this is typical in the generative AI space, where hosting models locally at scale is complex.

The code quality in the UI library is surprisingly clean for an open-source visual editor, with reusable React components and hooks managing node states and interactions. The backend follows FastAPI conventions, making it approachable for Python developers.

The Docker Compose setup simplifies deployment by bundling frontend, backend, and environment configurations, but requires Docker knowledge and may introduce some overhead compared to pure Node/Python local runs.

quick start

The README provides clear commands to get started either locally or with Docker Compose.

prerequisites

For local development, you need:

  • Node.js v20+
  • Python v3.10+
  • npm v7+ (for workspace support)

Or if you prefer Docker:

  • Docker v20+
  • Docker Compose v2+

installation and setup

# Clone the repo and install dependencies
git clone https://github.com/samuraigpt/vibe-workflow.git
cd vibe-workflow
npm install

configure backend

cd server
cp .env.example .env
# Then edit .env to set your MuAPI key:
# MU_API_KEY=your_actual_api_key_here

running frontend locally

npm run dev:app

running with docker compose

cp .env.example .env
# Edit .env to add your MuAPI key

docker compose up --build

Access the application at http://localhost:3000 and backend API at http://localhost:8000.

verdict

Vibe Workflow is a solid open-source option if you want a node-based AI workflow editor for generative image and video pipelines that you can self-host or embed. The modular UI library is the standout feature — few projects offer a reusable node editor component designed for integration into other React/Next.js apps.

It’s relevant for SaaS developers building AI content platforms who want to add a visual pipeline builder without reinventing the wheel or getting locked into proprietary services. Also useful for researchers and hobbyists experimenting with generative AI workflows.

The main limitation is the dependency on MuAPI for AI model execution, which requires a service API key and internet access. Running the backend and frontend locally requires juggling Node.js, Python, and environment configs, which might be a barrier for some.

Overall, the repo is well-structured, with clean code and good DX for both frontend and backend developers. If you need a flexible node editor for AI pipelines that can be embedded and customized, Vibe Workflow deserves a look.

→ GitHub Repo: SamurAIGPT/Vibe-Workflow ⭐ 362 · JavaScript