AI-native visual development environment for spaday

Overview

spaday-studio is a pilot AI-native visual development environment built on spaday's real component runtime. A structured project document is mirrored through transports; browser and MCP clients submit validated semantic edits; spaday applies the resulting tree diff without remounting unaffected elements.

The pilot includes:

• a selectable live application canvas with schema-driven property, binding, action, and state controls;
• component catalogs that retain Spaday property, field, event, and named-slot metadata;
• revision-checked title, key, property, binding, event, state, insert, move, and remove operations;
• private browser and MCP drafts with commit, discard, semantic rebase, conflict reporting, inspection, actor-aware undo and redo, and browser-refresh recovery;
• revision-scoped CodeMirror buffers using transports sequence CRDTs and cursor awareness;
• versioned atomic project persistence and deterministic Python export;
• host-defined read, edit, and admin roles for browser, WebSocket, and MCP operations;
• public JSON Schemas for documents, semantic operation batches, bindings, expressions, and actions;
• a persistent Spaday Store in the preview canvas, so authored bindings and actions run normally;
• a Python+JavaScript Copier scaffold, Playwright coverage, and Yardang/Sphinx documentation.

Run the pilot

make develop
make build
spaday-studio --project orbit.studio.json

Open http://127.0.0.1:8020. Select the headline, change its text or style, and preview the edit. The draft remains private until you click Commit. Accepted documents arrive through transports and the canvas applies a Spaday component-tree patch without replacing unaffected elements. Canonical edits are saved to orbit.studio.json. Use Export Python to download an ordinary Spaday page() function and its initial runtime state.

The MCP endpoint is http://127.0.0.1:8020/mcp. Start with the guided tutorial, connect an agent with the MCP how-to guide, save and export a project, set hosted access roles, use an installed component package, consult the API reference, or read why Studio uses a structured document.

Pilot boundaries

• Common field bindings and Store-writing actions have direct controls. Validated JSON editors retain the complete binding and action surface, plus runtime state.
• Concurrent leaf edits and anchor-based structural edits rebase when their targets remain valid. Numeric placement conflicts after another structural edit because its position is stale.
• Revision-scoped collaborative buffers are in-memory, pruned after their last draft retires, and are not part of the saved project.
• The default authorizer grants admin access for trusted local use. Network deployments must pass an authorizer and authenticate HTTP, WebSocket, and MCP connections upstream.
• Component packages are selected when the server starts; the browser cannot activate a package live.
• Arbitrary handwritten Python is not losslessly round-tripped.
• Python export is one-way generated output; the structured JSON project remains the editable source.

Note

This library was generated using copier from the Base Python Project Template repository.