Sekha - Universal AI Memory Controller

Universal AI memory controller that works across any LLM (OpenAI, Anthropic, Ollama). Built in Rust with dual storage (SQLite for metadata + ChromaDB for vectors), 17 REST endpoints, and native MCP protocol support. Features a 4-phase context assembly algorithm combining semantic search, recency, and importance scoring. Why we added it: Solves memory portability — your AI remembers conversations regardless of which model or tool you use.