FlowForge
A config-driven orchestration system for building composable AI workflows, where all node types are defined as JSON and rendered by a single universal controller.
FlowForge is a data-driven node abstraction layer that eliminates per-node component boilerplate by defining all node types as JSON config rendered through a single universal NodeController.
Traditional node-based workflow builders require per-node React components, creating scaling friction as new node types are added and making runtime extensibility difficult.
Engineered a type-based field registry (registerField/getField) mapping field type strings to React components at runtime, so new field types plug in without touching rendering logic. Dynamic ReactFlow handle generation auto-spawns input handles as users type {{variable}} syntax, parsed via regex diff against prior textarea state.
Type-based field registry over per-node components
WhyTraditional node-based workflow builders require a React component per node type, creating scaling friction as new node types are added and making runtime extensibility difficult.
AlternativeA dedicated React component per node type
ApproachA field registry (registerField/getField) maps field type strings to React components at runtime, so new field types plug in without touching rendering logic.
OutcomeZero per-node component boilerplate via a single universal NodeController.
Regex-diff parsing for dynamic handle generation
WhyWorkflow nodes need input handles that update live as a user types template variables.
ApproachDynamic ReactFlow handle generation auto-spawns input handles as users type {{variable}} syntax, parsed via regex diff against the prior textarea state.
OutcomeHandles stay in sync with node config with no manual wiring.
- →Zero per-node component boilerplate via universal NodeController
- →Runtime-extensible field registry — new field types require no rendering changes
- →Dynamic handle generation from {{variable}} syntax via regex diff parsing