AI/ML
Personal AI Memory: Cross-Platform MCP Server with DocumentDB Vector Search
A TypeScript MCP server that gives any AI assistant β Copilot CLI, Claude, Gemini β persistent personalized memory backed by DocumentDB. Uses a 4-layer retrieval strategy: cosmosSearch vector similarity, full-text search indexes, array tag queries, and regex fallback. Demonstrates DocumentDB's vector search (pgvector), MongoDB wire protocol compatibility, text indexes, atomic update operators, and flexible document schema in a practical AI application.