SQL retriever enhancements for Dell AIDp

industries/asset_lifecycle_management_agent/.cursor.rules.md

@@ -542,6 +542,120 @@ configs/
   .env                  # Environment variables
 ```
 
+## NAT Version Compatibility
+
+### NAT 1.2.1 vs 1.3.0
+
+**Current Version**: NAT 1.2.1 (with pydantic 2.10.x)
+
+**Key Compatibility Rules**:
+
+1. **Optional String Fields**:
+```python
+# ❌ WRONG - Will fail validation
+elasticsearch_url: str = Field(default=None)
+
+# ✅ CORRECT - Use Optional for nullable strings
+from typing import Optional
+elasticsearch_url: Optional[str] = Field(default=None)
+```
+
+2. **Reference Field Types (NAT 1.2.1)**:
+```python
+# NAT 1.2.1 uses plain strings for references
+llm_name: str = Field(description="LLM reference")
+embedding_name: str = Field(description="Embedder reference")
+```
+
+3. **Reference Field Types (NAT 1.3.0 - Future)**:
+```python
+# NAT 1.3.0 requires typed references
+from nat.data_models.component_ref import LLMRef, EmbedderRef, FunctionRef
+
+llm_name: LLMRef = Field(description="LLM reference")
+embedding_name: EmbedderRef = Field(description="Embedder reference")
+code_execution_tool: FunctionRef = Field(description="Function reference")
+```
+
+4. **YAML Configuration Quoting**:
+```yaml
+# Always quote string references in YAML configs for pydantic 2.10+
+functions:
+  sql_retriever:
+    llm_name: "sql_llm"              # Quoted
+    embedding_name: "vanna_embedder"  # Quoted
+    vector_store_type: "chromadb"     # Quoted
+    db_type: "sqlite"                 # Quoted
+
+  data_analysis_assistant:
+    tool_names: [
+      "sql_retriever",   # All tool names quoted
+      "predict_rul",
+      "plot_distribution"
+    ]
+```
+
+### Pydantic 2.10+ Best Practices
+
+**Type Annotations**:
+```python
+from typing import Optional
+
+class ToolConfig(FunctionBaseConfig):
+    # Required fields
+    required_param: str = Field(description="Must be provided")
+
+    # Optional fields with None default
+    optional_param: Optional[str] = Field(default=None, description="Can be None")
+
+    # Optional fields with non-None default
+    param_with_default: str = Field(default="default_value", description="Has default")
+
+    # Numeric fields (can use None without Optional if you want)
+    max_retries: int = Field(default=3, description="Number of retries")
+```
+
+**Common Validation Errors**:
+```
+ValidationError: Input should be a valid string [input_value=None, input_type=NoneType]
+→ Solution: Use Optional[str] instead of str for fields with default=None
+
+ValidationError: functions: Input should be a valid string (4 times)
+→ Solution: Quote all string values in YAML config, especially references
+```
+
+### Upgrading to NAT 1.3.0 (Future)
+
+When upgrading, you'll need to:
+
+1. Update pyproject.toml:
+```toml
+dependencies = [
+  "nvidia-nat[profiling,langchain,telemetry]==1.3.0",
+  "pydantic>=2.11.0,<3.0.0",
+]
+```
+
+2. Update all tool configs:
+```python
+# Before (NAT 1.2.1)
+llm_name: str = Field(...)
+
+# After (NAT 1.3.0)
+from nat.data_models.component_ref import LLMRef
+llm_name: LLMRef = Field(...)
+```
+
+3. Update evaluator configs:
+```python
+# multimodal_llm_judge_evaluator_register.py
+# llm_judge_evaluator_register.py
+from nat.data_models.component_ref import LLMRef
+llm_name: LLMRef = Field(...)
+```
+
+4. Keep Optional[str] for nullable fields (both versions need this)
+
 ## Debugging and Troubleshooting
 
 ### Common Issues and Solutions

industries/asset_lifecycle_management_agent/.gitignore

@@ -1,46 +1,123 @@
-# macOS system files
+# Misc
+config_examples.yml
+config_examples.yaml
+env.sh
+frontend/
+prompts.md
+
+# Python
+__pycache__/
+*.py[cod]
+*$py.class
+*.so
+.Python
+*.egg
+*.egg-info/
+dist/
+build/
+*.whl
+pip-wheel-metadata/
 .DS_Store
-.DS_Store?
-._*
-.Spotlight-V100
-.Trashes
-ehthumbs.db
-Thumbs.db
 
-# Database and vector store files
-database/
-*.db
-*.sqlite3
+# Virtual environments
+.venv/
+venv/
+ENV/
+env/
+
+# IDEs and Editors
+.vscode/
+.idea/
+*.swp
+*.swo
+*~
+.DS_Store
+
+# Testing
+.pytest_cache/
+.coverage
+htmlcov/
+.tox/
+.hypothesis/
+
+# Jupyter Notebook
+.ipynb_checkpoints/
+*.ipynb_checkpoints/
 
-# Output and generated files
+# Output and Data Directories
 output_data/
-moment/
-readmes/
-*.html
-*.csv
-*.npy
+eval_output/
+example_eval_output/
+output/
+results/
+logs/
 
-# Python package metadata
-src/**/*.egg-info/
-*.egg-info/
+# Database files
+*.db
+*.sqlite
+*.sqlite3
+database/*.db
+database/*.sqlite
 
-# Environment files (if they contain secrets)
-env.sh
+# Vector store data (ChromaDB)
+database/
+chroma_db/
+vector_store/
+vanna_vector_store/
 
-# Model files (if large/binary)
+# Model files (large binary files)
 models/*.pkl
-models/*.joblib
-models/*.model
+models/*.h5
+models/*.pt
+models/*.pth
+models/*.ckpt
+*.pkl
+*.h5
+*.pt
+*.pth
+moment/
 
-# Logs
-*.log
-logs/
+# Data files (CSV, JSON, etc. - be selective)
+*.csv
+*.json
+!training_data.json
+!vanna_training_data.yaml
+!config*.json
+!config*.yaml
+!config*.yml
+!pyproject.toml
+!package.json
+
+# Frontend build artifacts
+frontend/node_modules/
+frontend/dist/
+frontend/build/
+frontend/.next/
+frontend/out/
+
+# Environment and secrets
+.env
+.env.local
+.env.*.local
+*.secret
+secrets/
+credentials/
 
 # Temporary files
 *.tmp
 *.temp
-.pytest_cache/
-__pycache__/ 
+*.log
+*.cache
+
+# OS specific
+Thumbs.db
+Desktop.ini
+
+# Experiment tracking
+mlruns/
+wandb/
 
-# dot env
-mydot.env
+# Documentation builds
+docs/_build/
+docs/.doctrees/
+site/