How chanduchowday25/agent-os works
AgentOS is a multi-agent workspace for document-grounded chat, analytics, and reports. Users interact through the React application; its chat service calls the FastAPI API, while the backend’s agent workflow coordinates specialized agents. Document Q&A stages include upload, extraction and chunking, embedding and indexing, hybrid retrieval, relevance gating, and grounded response with citations. The sampled source confirms ingestion, retrieval, embedding, and RAG internals; route-to-service wiring for uploads, analytics, reports, and persistence was not sampled, so those capabilities are shown without speculative connections.
Main entry points: router.py API router · App.jsx React app · retrieval_service.py Hybrid retrieval · embedding_service.py Embedding service
Components 25
User workspace
- React app
agentos-scaffold/frontend/src/App.jsx - Workspace
agentos-scaffold/frontend/src/pages/Workspace.jsx - Analytics page
agentos-scaffold/frontend/src/pages/Analytics.jsx - Reports page
agentos-scaffold/frontend/src/pages/Reports.jsx - Chat state
agentos-scaffold/frontend/src/hooks/useChat.js - Chat API client
agentos-scaffold/frontend/src/services/chatService.js
API surface
- API router
agentos-scaffold/backend/api/v1/router.py - Chat endpoint
agentos-scaffold/backend/api/v1/routes/chat.py - Upload endpoint
agentos-scaffold/backend/api/v1/routes/upload.py - Analysis endpoint
agentos-scaffold/backend/api/v1/routes/analyze.py - Reports endpoint
agentos-scaffold/backend/api/v1/routes/reports.py - Conversation endpoints
agentos-scaffold/backend/api/v1/routes/conversations.py
Agent orchestration
- Agent workflow
agentos-scaffold/backend/workflows/agent_workflow.py - Supervisor agent
agentos-scaffold/backend/agents/supervisor_agent.py - RAG agent
agentos-scaffold/backend/agents/rag_agent.py - Analytics agent
agentos-scaffold/backend/agents/analytics_agent.py - Research agent
agentos-scaffold/backend/agents/research_agent.py
Document intelligence
- Document ingestion
agentos-scaffold/backend/services/document_ingestion_service.py - Hybrid retrieval
agentos-scaffold/backend/services/retrieval_service.py - Vector index
agentos-scaffold/backend/database/chroma_client.py
Shared services
- Embedding service
agentos-scaffold/backend/services/embedding_service.py - LLM service
agentos-scaffold/backend/services/llm_service.py
Other components
- User
- Gemini embeddings
- Groq model API
Connections 20
- User → React app: uses
- React app → Workspace: routes to
- React app → Analytics page: routes to
- React app → Reports page: routes to
- Workspace → Chat state: uses
- Chat state → Chat API client: calls
- Chat API client → Chat endpoint: HTTP request
- API router → Chat endpoint: dispatches
- API router → Upload endpoint: dispatches
- API router → Analysis endpoint: dispatches
- API router → Reports endpoint: dispatches
- API router → Conversation endpoints: dispatches
- Document ingestion → Embedding service: embeds chunks
- Document ingestion → Vector index: indexes chunks
- Hybrid retrieval → Embedding service: embeds query
- Hybrid retrieval → Vector index: searches index
- RAG agent → Hybrid retrieval: retrieves context
- RAG agent → LLM service: generates answer
- Embedding service → Gemini embeddings: requests embeddings
- LLM service → Groq model API: requests completion