How kinginthenorth12/agentic_ai_frontforge works
FrontForge AI turns a user’s website request into a locally generated React project. The main workflow gathers missing requirements, plans the project, creates a normalized UI blueprint, then generates its components and entrypoints. For generation, the component agent retrieves relevant reference documentation and writes code into the project. Documentation ingestion prepares that local retrieval store. The graph reflects the sampled entrypoints; some implementation details are omitted from the excerpts.
Main entry points: component_agent.py Component generator · main.py Workflow runner · UI_Architect.py UI architect · clarification_agent.py Requirement clarification
Components 13
Generation workflow
- Workflow runner
main.py
Requirements and planning
- Requirement clarification
clarification_agent.py - Project planner
planner_agent.py - Project plan
Blueprint and code generation
- UI architect
UI_Architect.py - UI blueprint
- Component generator
component_agent.py - Generated React project
Local reference knowledge
- Documentation ingestion
ingest.py - Reference retriever
retriever.py - Chroma reference store
Other components
- User
- Local Ollama models
Connections 20
- User → Requirement clarification: provides request
- Workflow runner → Requirement clarification: calls
- Requirement clarification → Local Ollama models: invokes model
- Requirement clarification → Workflow runner: returns requirements
- Workflow runner → Project planner: calls
- Project planner → Local Ollama models: invokes model
- Project planner → Project plan: writes plan
- Workflow runner → UI architect: calls
- UI architect → Project plan: reads plan
- UI architect → Local Ollama models: invokes model
- UI architect → UI blueprint: writes blueprint
- Workflow runner → Component generator: calls
- Component generator → UI blueprint: reads blueprint
- Component generator → Project plan: reads plan
- Component generator → Reference retriever: queries
- Reference retriever → Chroma reference store: searches
- Component generator → Local Ollama models: invokes model
- Component generator → Generated React project: writes generated code
- Documentation ingestion → Local Ollama models: embeds documents
- Documentation ingestion → Chroma reference store: stores chunks