v0.3.0
bytedance/flowgram.aiv0.3.0Jul 30, 2025by xiamidaxia
AI Summary
This major release (v0.3.0) introduces a significant architectural refactoring with the TypeManager system to centrally manage type-related logic, replacing the previous approach that required modifying multiple materials when adding new types. The release also adds a new @flowgram.ai/json-schema package, improves JSON input for Object and Array types, and provides DisplayInputsValues and DisplayOutputs components for auto-refreshing variable types in node cards.
Key Highlights
- Introduced TypeManager for unified type management through plugins instead of modifying multiple materials
- New @flowgram.ai/json-schema package for JSON schema materials
- Added createTypePresetPlugin for plugin-based type definitions
- Improved Constant JSON Input for Object and Array types
- New DisplayInputsValues and DisplayOutputs components for auto-refreshing variable types in node cards
Breaking Changes
- Type management architecture has been refactored - types are now managed through TypeManager plugins rather than direct material modifications
New Features
- TypeManager system for centralized type logic management
- @flowgram.ai/json-schema package
- createTypePresetPlugin for defining custom types via plugins
- Enhanced JSON Input component for Object and Array types
- DisplayInputsValues component for displaying input values with auto type refresh
- DisplayOutputs component with displayFromScope for fetching variables from variable engine
Full Release Notes
## What's Changed
* feat(variable): add @flowgram.ai/json-schema and adjust json schema's materials by @sanmaopep in https://github.com/bytedance/flowgram.ai/pull/602
### Introduce TypeManager to manage type-related logics
We refactored the materials related to types.
All type logic is converged through `typeManager` to prevent changes to types from spreading to a huge amout of materials.
- In version 0.2, when we added a new type, we needed to modify multiple materials such as `ConstantInput` and `TypeSelector`, causing a domino effect.
- In version 0.3, unified management can be achieved through plugins:
```typescript
import { createTypePresetPlugin } from '@flowgram.ai/form-materials'
createTypePresetPlugin({
types: [
{
type: 'file',
label: 'File',
icon: <svg>xxxx</svg>,
ConstantRenderer: () => {
// ...
},
},
],
});
```
### Constant Json Input for Object and Array Type
We update Json Input for Object and Array Type
<img width="481" height="126" alt="image" src="https://github.com/user-attachments/assets/9aabc4c6-7303-4ebd-aed5-724f8c41d965" />
### `DisplayInputsValues` and `DisplayOutputs`
We provide `DisplayInputsValues` and `DisplayOutputs`, which will auto refresh variable type in node card
```typescript
// variable ref type will auto refreshed with this material
<Field<Record<string, IFlowValue | undefined> | undefined> name="inputsValues">
{({ field }) => <DisplayInputsValues value={field.value} />}
</Field>
// displayFromScope will auto fetch variables information from variable engine in current scope
<DisplayOutputs displayFromScope />
```
Here is the demo:
**Full Changelog**: https://github.com/bytedance/flowgram.ai/compare/v0.2.32...v0.3.0