wahhabriaz/agentic-document-rag

How wahhabriaz/agentic-document-rag works

HybridRAG is a PDF assistant: users upload documents, then ask questions answered from both retrieved text and extracted tables. The graph follows the documented upload and Q&A flows through the FastAPI backend, processing/storage stages, and agent system, while retaining the Streamlit client and data-clearing capability. Some source excerpts are partial; uncertain internal dispatch is shown as unverified rather than asserted.

Main entry points: chat.py Chat API · manager_agent.py Manager Agent · clear_data_service.py Clear Data Service · upload_pdf.py Upload Handler

Components 20

User Interface

Streamlit UI src/frontend/streamlit_app.py

API and Services

FastAPI App src/backend/__init__.py
Chat API src/backend/routes/chat.py
API Models src/backend/models.py
Configuration src/backend/config.py
Query Orchestrator src/backend/services/orchestrator.py
Clear Data Service src/backend/services/clear_data_service.py
Clear Data CLI clear_data_script.py

PDF Ingestion

Upload Handler src/backend/utils/upload_pdf.py
PDF Processor src/backend/utils/pdf_processor.py
Schema Manager src/backend/utils/schema_manager.py
Embedding Service src/backend/services/embedding_service.py

Answering Agents

Manager Agent src/backend/agents/manager_agent.py
RAG Agent src/backend/agents/rag_agent.py
Table Agent src/backend/agents/table_agent.py
Response Combiner src/backend/agents/combiner_agent.py
Agent Base src/backend/agents/base.py

Data Stores

Pinecone Index
MySQL Tables

Other components

User

Connections 25

User → Streamlit UI: uses README.md
Streamlit UI → Chat API: calls API README.md
FastAPI App → Chat API: registers src/backend/__init__.py
Chat API → API Models: validates src/backend/routes/chat.py
Chat API → Upload Handler: dispatches upload src/backend/routes/chat.py
Upload Handler → PDF Processor: extracts content src/backend/utils/upload_pdf.py
Upload Handler → Embedding Service: stores embeddings src/backend/utils/upload_pdf.py
PDF Processor → MySQL Tables: stores tables README.md
Embedding Service → Pinecone Index: stores vectors src/backend/services/embedding_service.py
PDF Processor → Schema Manager: manages schemas
Chat API → Query Orchestrator: submits query src/backend/routes/chat.py
Query Orchestrator → Manager Agent: dispatches query
Manager Agent → RAG Agent: runs retrieval src/backend/agents/manager_agent.py
Manager Agent → Table Agent: runs table query src/backend/agents/manager_agent.py
Manager Agent → Response Combiner: combines answers src/backend/agents/manager_agent.py
RAG Agent → Pinecone Index: searches vectors src/backend/agents/rag_agent.py
Table Agent → MySQL Tables: queries tables README.md
RAG Agent → Agent Base: extends src/backend/agents/rag_agent.py
FastAPI App → Query Orchestrator: initializes src/backend/__init__.py
Chat API → Clear Data Service: clears data src/backend/routes/chat.py
Clear Data Service → Pinecone Index: deletes vectors src/backend/services/clear_data_service.py
Clear Data Service → MySQL Tables: drops tables src/backend/services/clear_data_service.py
Clear Data CLI → Pinecone Index: clears vectors clear_data_script.py
Clear Data CLI → MySQL Tables: clears tables clear_data_script.py
Configuration → FastAPI App: configures src/backend/__init__.py

Drawn by GitDiagram on November 12, 2026 from the repository's file tree, README and source files. Read as Markdown ·