feat: Add enhanced configuration system with multi-provider LLM support
This commit introduces a comprehensive configuration system that makes
Graphiti more flexible and easier to configure across different
providers and deployment environments.
## New Features
- **Unified Configuration**: New GraphitiConfig class with Pydantic validation
- **YAML Support**: Load configuration from .graphiti.yaml files
- **Multi-Provider Support**: Easy switching between OpenAI, Azure, Anthropic,
Gemini, Groq, and LiteLLM
- **LiteLLM Integration**: Unified access to 100+ LLM providers
- **Factory Functions**: Automatic client creation from configuration
- **Full Backward Compatibility**: Existing code continues to work
## Configuration System
- graphiti_core/config/settings.py: Pydantic configuration classes
- graphiti_core/config/providers.py: Provider enumerations and defaults
- graphiti_core/config/factory.py: Factory functions for client creation
## LiteLLM Client
- graphiti_core/llm_client/litellm_client.py: New unified LLM client
- Support for Azure OpenAI, AWS Bedrock, Vertex AI, Ollama, vLLM, etc.
- Automatic structured output detection
## Documentation
- docs/CONFIGURATION.md: Comprehensive configuration guide
- examples/graphiti_config_example.yaml: Example configurations
- DOMAIN_AGNOSTIC_IMPROVEMENT_PLAN.md: Future improvement roadmap
## Tests
- tests/config/test_settings.py: 22 tests for configuration
- tests/config/test_factory.py: 12 tests for factories
- 33/34 tests passing (97%)
## Issues Addressed
- #1004: Azure OpenAI support
- #1006: Azure OpenAI reranker support
- #1007: vLLM/OpenAI-compatible provider stability
- #1074: Ollama embeddings support
- #995: Docker Azure OpenAI support
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
74a422369c