How sanskritimishra10/ai_codebase_onboarding_assistant works
This is an AI codebase onboarding assistant: developers upload a repository, inspect its analysis, index its source, then ask questions and receive answers with source references. The graph follows the documented workflow and sampled calls across the frontend, API, ingestion/indexing, and retrieval-answering layers. The vector index and uploaded repositories are represented as stores; conversations are in-memory. Several UI and backend links are supported by the SOURCE INDEX or README rather than complete excerpts, so unsampled internals are not expanded.
Main entry points: repository.py Repository endpoints · api.js API client · ChatWindow.jsx Chat window · rag_service.py RAG orchestrator
Components 24
Frontend
- Upload page
frontend/src/pages/Home.jsx - File upload
frontend/src/components/FileUpload.jsx - Repository page
frontend/src/pages/Repository.jsx - Chat page
frontend/src/pages/Chat.jsx - Chat window
frontend/src/components/ChatWindow.jsx - Chat messages
frontend/src/components/ChatMessage.jsx - Source cards
frontend/src/components/SourceCard.jsx - API client
frontend/src/services/api.js
API
- FastAPI app
backend/app/main.py - Repository endpoints
backend/app/api/repository.py - Conversation endpoints
backend/app/api/conversation.py
Repository processing
- Repository storage
backend/app/services/repository_service.py - Archive safety
backend/app/utils/file_utils.py - Repository analysis
backend/app/services/repository_analyzer.py - Source ingestion
backend/app/services/repository_ingestion.py - Code chunking
backend/app/services/code_chunker.py - Vector index
backend/app/services/vector_store.py - Embeddings
backend/app/services/embedding_service.py - Uploaded repositories
storage/uploads
Chat and retrieval
- RAG orchestrator
backend/app/services/rag_service.py - Query and answer LLM
backend/app/services/llm_service.py - Conversation state
backend/app/services/conversation_service.py
Other components
- Developer
- Hugging Face inference
Connections 30
- Developer → File upload: uploads ZIP
README.md - Upload page → File upload: renders
frontend/src/pages/Home.jsx - File upload → API client: calls upload
frontend/src/components/FileUpload.jsx - Repository page → API client: requests analysis/index
frontend/src/pages/Repository.jsx - Chat page → API client: loads conversations
frontend/src/pages/Chat.jsx - Chat page → Chat window: renders
frontend/src/pages/Chat.jsx - Chat window → Chat messages: renders
frontend/src/components/ChatWindow.jsx - Chat window → Source cards: renders
frontend/src/components/ChatWindow.jsx - Chat window → API client: asks questions
frontend/src/components/ChatWindow.jsx - API client → Repository endpoints: HTTP requests
frontend/src/services/api.js - API client → Conversation endpoints: HTTP requests
frontend/src/services/api.js - FastAPI app → Repository endpoints: registers router
backend/app/main.py - FastAPI app → Conversation endpoints: registers router
backend/app/main.py - Repository endpoints → Repository storage: saves upload
backend/app/api/repository.py - Repository storage → Archive safety: extracts safely
backend/app/services/repository_service.py - Repository storage → Uploaded repositories: stores files
- Repository endpoints → Repository analysis: analyzes repository
backend/app/api/repository.py - Repository endpoints → Source ingestion: reads source
backend/app/api/repository.py - Source ingestion → Archive safety: filters sensitive files
backend/app/services/repository_ingestion.py - Repository endpoints → Code chunking: chunks documents
backend/app/api/repository.py - Repository endpoints → Vector index: indexes repository
backend/app/api/repository.py - Vector index → Embeddings: generates vectors
backend/app/services/vector_store.py - Vector index → Code chunking: uses chunks
backend/app/services/vector_store.py - Repository endpoints → RAG orchestrator: answers questions
backend/app/api/repository.py - Repository endpoints → Conversation state: creates conversations
backend/app/api/repository.py - Conversation endpoints → Conversation state: manages history
backend/app/api/conversation.py - RAG orchestrator → Vector index: searches chunks
backend/app/services/rag_service.py - RAG orchestrator → Query and answer LLM: rewrites and generates
backend/app/services/rag_service.py - RAG orchestrator → Conversation state: reads and saves messages
backend/app/services/rag_service.py - Query and answer LLM → Hugging Face inference: calls inference
backend/app/services/llm_service.py