Xuda
Reactive runtime and app tooling for real products
Ship reactive apps that AI can read, write, and reason about.
Xuda gives you a stable, predictable app structure that both humans and AI models can understand from day one. Use Xuda Slim for HTML-first reactive pages, or Xuda CLI for full source-driven applications with typed folders, generated runtime docs, and optional cloud sync.
React, Vue, and vanilla JS were built for humans writing code. Xuda is built so AI can understand the entire app.
Traditional frameworks give you maximum flexibility, but that flexibility is the problem for AI. An AI model looking at a React or Vue project sees thousands of possible patterns, custom folder structures, ad-hoc state management, and no single canonical way to represent what the app does. Xuda takes the opposite approach: one stable structure, explicit building blocks, and a generated runtime doc that describes the full app in a single readable format.
Every project follows one structure.
components/, data/get/, data/set/, alerts/, tables/, routes/.
get_data, set_data, batch — typed, named, predictable.
An AI model that knows Xuda can build any Xuda app. That is not true for React or Vue.
When you ask an AI to build a React app, it has to make dozens of architectural decisions first: which router, which state library, which data layer, which folder layout. The result works, but it will not match your team's existing patterns. With Xuda, the decisions are already made. The AI reads the technical docs once and produces apps that look like every other Xuda app — because there is only one way to structure them.
Stable source layout
Predictable folders like components/, globals/, data/get/, and alerts/ mean agents can locate intent without scanning the whole repo.
11 explicit building blocks
Every concern has a named type: component, get_data, set_data, batch, api, alert, javascript, table, route, globals, ai_agent.
One compiled runtime doc
Source compiles into a single canonical JSON. An AI does not need to trace imports across files — everything is in one place.
Published technical docs
xuda.io/technical-docs publishes HTML and raw markdown for every concept, so AI systems and developers learn from the same material.
One runtime, two authoring modes
Xuda Slim and Xuda CLI sit on the same reactive model. An AI assistant can move between lightweight pages and larger apps without relearning anything.
No framework soup
No choosing between Next, Nuxt, Remix, Vite plugins, or meta-framework conventions. Xuda is the framework, the build tool, and the runtime in one package.
Less tokens, less energy, less risk. More shipping.
When AI understands the entire app structure upfront, everything downstream gets cheaper, faster, and safer. These are not abstract advantages — they show up on every prompt, every generation, every deploy.
Dramatically fewer tokens
A React project forces the AI to read dozens of files to understand the architecture. Xuda compiles the full app into one runtime doc — the AI reads one file, uses fewer input tokens, and responds faster. Less context means lower cost per prompt.
Lower energy consumption
Every token an AI processes costs compute and electricity. Fewer tokens per interaction, fewer retries from misunderstood architecture, and fewer wasted generation cycles add up. A structured platform is literally a greener one.
Reduced code poisoning risk
When an AI has to guess your project structure, it may introduce patterns from other codebases it was trained on — wrong dependencies, unsafe defaults, or stale APIs. Xuda's fixed structure eliminates this: the AI generates within a known, bounded system.
No hallucinated architecture
AI models hallucinate most when they lack constraints. Ask for a "Vue dashboard" and you get an invented folder layout. Ask for a "Xuda dashboard" and the AI follows the one documented structure. Fixed constraints kill hallucination.
Consistent output across models
GPT-4, Claude, Gemini, or a local model — any AI that reads the Xuda technical docs will produce the same app shape. You are not locked to one model's interpretation of "best practices." The platform defines best practices.
Fewer review cycles
When AI-generated code matches your existing project structure every time, code review shrinks to logic review. No more "why did it use Redux here" or "this doesn't follow our conventions." The conventions are enforced by the platform itself.
Create a new Xuda app, install dependencies, and run it locally in a few commands.
The recommended starting point is Xuda CLI. It gives you the root app file, typed source folders, generated runtime docs, and the local development server out of the box.
Download and run
npm create xuda-cli@latest my-app cd my-app npm install npm run dev
This creates a new app, installs the published packages, and starts the local Xuda development server.
Try a full example app
mkdir xuda-studant-app-example cd xuda-studant-app-example npm pack xuda-studant-app-example tar -xzf xuda-studant-app-example-*.tgz --strip-components=1 package rm xuda-studant-app-example-*.tgz npm install npm run dev
This pulls down the published student dashboard example as a real app folder, then installs and runs it in place. Use this path when you want to inspect a larger Xuda app with components, tables, routes, alerts, APIs, and database-backed examples.
Do not install the example into another app with plain npm install xuda-studant-app-example
and expect a local dev script to appear. The package is an app template, not a library.
What you get
index.html as the document shell
App.xu as the root Xuda screen
Xuda is the reactive technology layer. Slim and CLI are two ways to author against it.
The same Xuda runtime ideas show up in both products: datasource-backed fields, expressions, events, composable programs, and reusable building blocks. The difference is how much structure you want around authoring, generation, and app lifecycle.
Xuda Slim
Use Slim when you want HTML-first reactive pages, embeddable widgets, lightweight product surfaces, or documentation examples that stay visually close to the DOM.
Xuda CLI
Use Xuda CLI when you want a source-first app structure with App.xu, typed folders such as
components/ and data/get/, generated runtime docs, and optional cloud sync.
Choose the lighter path for pages and the fuller path for applications.
Xuda Slim and Xuda CLI are complementary. The first is ideal when you want to stay close to hand-authored HTML. The second is better when you want application structure, source files per building block, and a clear compile-and-runtime story.
Choose this when the page is the product surface.
Great for marketing pages, embedded tools, docs demos, or any place where HTML is the natural source of truth and you want Xuda reactivity without a larger framework.
Choose this when the app itself needs source architecture.
Great for dashboards, panels, form-driven apps, multi-screen products, datasource-backed workflows, and
teams that want reusable source units such as get_data, set_data, and
alert.
Move from overview into examples, docs, or technical implementation guidance.
Use this site for the product-level story, examples, and navigation. Use
xuda.io/technical-docs for the crawler-friendly technical reference that helps developers and
AI systems build with Xuda from scratch.
Compare paths
See a simple decision guide for when to use Xuda Slim versus Xuda CLI.
Stay in Slim
Read the current Slim docs and explore the HTML and JSON examples already on this site.
Go technical
Read the technical docs for both Xuda Slim and Xuda CLI, with raw markdown published for crawlers and AI tools.