How shirsadip0907/gen_ai works
Main system components
• Frontend/UI: The interactive user interface built using Streamlit (represented by app.py).
• Multi-Agent Ecosystem: The pool of specialized agents – Project Lead, Business Analyst, Designer, Developer, and Tester (found under the agents/ directory). Each of these agents is responsible for a specific phase of the SDLC.
• Data and Artifact Storage: The ChromaDB component and associated SQLite databases (directories under chroma_db and the knowledge_base under data/knowledge_base).
• Database Management: The database module (database/db_manager.py) for artifact storage, retrieval, and versioning.
• Utility Module: Helper functions and tools (under utils/).
Interactions and data flows
• The user interacts with the Frontend (Streamlit), triggering project initialization.
• The Project Lead agent processes the project initiation and coordinates subsequent phases.
• The Business Analyst agent takes requirements and converts them into detailed user stories.
• The Designer agent generates the system and architectural design documents.
• The Developer agent converts these specifications into functional code through iterative feedback.
• The Tester agent is responsible for quality assurance by running tests and validating the final product.
• Artifacts and outputs from each phase are stored and retrieved from ChromaDB, ensuring continuity and context transfer throughout the lifecycle.
• The database manager acts as the custodian of these artifacts, interfacing with ChromaDB.
Architectural patterns and technology stacks
• Layered structure (Frontend, Agent orchestration, Data management, and Utilities).
• The overall pattern resembles a workflow pipeline, where each phase produces outputs that serve as inputs to the next.
• Noted design principles include separation of concerns, encapsulation (each AI agent has a defined role), and context propagation (via artifact storage and retrieval leveraging RAG capabilities).
• Technologies in use: Python 3.9+, Streamlit (frontend), ChromaDB/SQLite (data storage), LangChain and LiteLLM (LLM integrations), and the multi-agent framework CrewAI.