How gaurav-nova1702/aria works
NeuralMind is a React/FastAPI self-learning agent. A user signs in, sends a chat message through the frontend API to the backend chat endpoint, and the orchestrator enriches generation with memory, knowledge, research and positive feedback examples before returning a response. Users can also search and save knowledge, run research, rate answers, use browser speech, and view metrics. The source excerpts establish several backend calls; other links below are supported by the README or SOURCE INDEX, so unsampled internals are not inferred.
Main entry points: client.js API client · orchestrator.py Agent orchestrator · App.jsx App shell · knowledge.py Knowledge manager
Components 28
Frontend experience
- App shell
frontend/src/App.jsx - Chat interface
frontend/src/components/ChatWindow.jsx - Chat state hook
frontend/src/hooks/useChat.js - Agent state
frontend/src/store/agentStore.js - Voice interaction
frontend/src/hooks/useVoice.js - Research panel
frontend/src/components/ResearchPanel.jsx - Knowledge browser
frontend/src/components/KnowledgeBase.jsx - Metrics dashboard
frontend/src/components/MetricsPanel.jsx - Feedback controls
frontend/src/components/FeedbackButtons.jsx
API and identity
- API client
frontend/src/api/client.js - Token verification
backend/utils/auth.py - Chat endpoints
backend/api/chat.py - Knowledge endpoints
backend/api/knowledge.py - Research endpoint
backend/api/research.py - Feedback endpoints
backend/api/feedback.py
Agent capabilities
- Agent orchestrator
backend/agent/orchestrator.py - Conversation memory
backend/agent/memory.py - Knowledge manager
backend/agent/knowledge.py - Document processing
backend/agent/document_processor.py - Web research
backend/agent/research.py - Feedback learning
backend/agent/feedback.py
Persistence and retrieval
- Embedding service
backend/utils/embeddings.py - Vector storage
backend/agent/pgvector_store.py - Relational database
backend/models/database.py
Other components
- User
- Google Gemini
- Supabase Auth
- Web search and pages
Connections 39
- User → App shell: uses
- App shell → Chat interface: renders
frontend/src/App.jsx - App shell → Research panel: renders
frontend/src/App.jsx - App shell → Knowledge browser: renders
frontend/src/App.jsx - App shell → Metrics dashboard: renders
frontend/src/App.jsx - Chat interface → Chat state hook: uses
- Chat state hook → API client: calls
frontend/src/hooks/useChat.js - Chat state hook → Agent state: updates
frontend/src/hooks/useChat.js - Voice interaction → Agent state: updates
frontend/src/hooks/useVoice.js - Research panel → API client: searches through
frontend/src/components/ResearchPanel.jsx - Feedback controls → API client: submits through
- Knowledge browser → API client: requests through
- Metrics dashboard → API client: requests through
- API client → Supabase Auth: gets session token
frontend/src/api/client.js - API client → Chat endpoints: calls REST
- API client → Knowledge endpoints: calls REST
- API client → Research endpoint: calls REST
- API client → Feedback endpoints: calls REST
- Chat endpoints → Token verification: verifies identity
backend/api/chat.py - Knowledge endpoints → Token verification: verifies identity
backend/api/knowledge.py - Research endpoint → Token verification: verifies identity
backend/api/research.py - Chat endpoints → Agent orchestrator: dispatches chat
backend/api/chat.py - Knowledge endpoints → Document processing: processes uploads
backend/api/knowledge.py - Knowledge endpoints → Knowledge manager: manages entries
backend/api/knowledge.py - Research endpoint → Web research: runs search
backend/api/research.py - Feedback endpoints → Feedback learning: handles ratings
- Agent orchestrator → Conversation memory: retrieves history
backend/agent/orchestrator.py - Agent orchestrator → Knowledge manager: searches knowledge
backend/agent/orchestrator.py - Agent orchestrator → Web research: requests research
backend/agent/orchestrator.py - Agent orchestrator → Feedback learning: gets examples
backend/agent/orchestrator.py - Agent orchestrator → Google Gemini: generates response
- Knowledge manager → Embedding service: creates vectors
backend/agent/knowledge.py - Knowledge manager → Vector storage: searches vectors
backend/agent/knowledge.py - Knowledge manager → Relational database: persists entries
backend/agent/knowledge.py - Web research → Web search and pages: searches and fetches
backend/agent/research.py - Web research → Google Gemini: summarizes pages
backend/agent/research.py - Web research → Relational database: caches results
backend/agent/research.py - Feedback learning → Relational database: reads ratings
backend/agent/feedback.py - Conversation memory → Relational database: stores exchanges