Add CLAUDE.md (#156)

Add comprehensive guidance for CLAUDE development, including setup, code quality, architecture, and dataset information

Signed-off-by: Jack Luar <jluar@precisioninno.com>

Author: luarss

CLAUDE.md

@@ -0,0 +1,124 @@
+# CLAUDE.md
+
+This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
+
+## Commands
+
+### Development Environment Setup
+```bash
+# Initialize all components with dev dependencies
+make init-dev
+
+# Initialize individual components
+cd backend && make init-dev
+cd frontend && make init-dev
+cd evaluation && make init-dev
+```
+
+### Code Quality
+```bash
+# Format all code
+make format
+
+# Run all checks (linting, type checking, pre-commit)
+make check
+
+# Individual component checks
+cd backend && make check  # mypy + ruff
+cd frontend && make check  # mypy + ruff  
+cd evaluation && make check  # mypy + ruff
+```
+
+### Backend Development
+```bash
+# Run the FastAPI server
+cd backend && python main.py
+
+# Run terminal chatbot
+cd backend && python chatbot.py
+
+# Build documentation dataset
+cd backend && python build_docs.py
+```
+
+### Frontend Development
+```bash
+# Run Streamlit app
+cd frontend && streamlit run streamlit_app.py
+```
+
+### Evaluation
+```bash
+# Run LLM evaluation tests
+cd evaluation && make llm-tests
+```
+
+### Docker
+```bash
+# Start all services
+make docker-up
+
+# Stop all services
+make docker-down
+```
+
+## Architecture
+
+This is a RAG-based conversational assistant for OpenROAD with three main components:
+
+### Backend (`backend/`)
+- **FastAPI server** with chat endpoints
+- **LangGraph-based routing** - Routes queries to specialized retrievers based on query classification
+- **Multiple retriever tools** for different content types:
+  - General OR/ORFS information
+  - OR tools and commands  
+  - Installation docs
+  - Error messages
+  - OpenSTA docs
+  - Yosys docs
+  - Klayout docs
+- **FAISS vector databases** for document retrieval
+- **Hybrid retrieval** combining semantic search, MMR, and BM25
+- **Cross-encoder reranking** for improved relevance
+
+### Frontend (`frontend/`)
+- **Streamlit-based chat interface**
+- **User feedback collection** via Google Sheets API
+- **MongoDB integration** for data persistence
+
+### Evaluation (`evaluation/`)
+- **Automated LLM evaluation** using various metrics
+- **Human evaluation interface** for research
+- **Performance benchmarking** tools
+
+## Key Files
+
+- `backend/src/agents/retriever_graph.py` - Main LangGraph routing logic
+- `backend/src/chains/` - Individual retriever implementations
+- `backend/build_docs.py` - Documentation dataset builder
+- `backend/src/vectorstores/faiss.py` - Vector database implementation
+- `frontend/streamlit_app.py` - Main Streamlit interface
+- `evaluation/auto_evaluation/eval_main.py` - Automated evaluation runner
+
+## Development Guidelines
+
+- Use **Python 3.12** with virtual environments
+- Follow **ruff** formatting and linting standards
+- Run **mypy** for type checking
+- Use **pre-commit hooks** for code quality
+- Components are designed to be **independently deployable**
+
+## Dataset
+
+The RAG dataset is hosted on HuggingFace and contains:
+- OpenROAD documentation
+- OpenROAD-flow-scripts documentation  
+- OpenSTA documentation
+- Yosys documentation
+- Klayout documentation
+- GitHub discussions and issues
+
+Download with:
+```bash
+huggingface-cli download The-OpenROAD-Project/ORAssistant_RAG_Dataset --repo-type dataset --local-dir data/
+```