kg-gen converts unstructured text into structured knowledge graphs using LLMs. It supports multiple model providers (OpenAI, Anthropic, Gemini, Ollama, Deepseek) through LiteLLM and uses DSPy for structured output generation. The tool is designed for RAG assistance, synthetic data creation for ML training, and relationship analysis in source text. It ships with an MCP server for integration into AI tooling ecosystems.