Context Portal MCP (ConPort) is a memory-bank backend that stores structured project context in a per-workspace SQLite database and exposes an MCP API built with Python and FastAPI. It serves AI assistants and developer tools by enabling storage, retrieval, and management of diverse project data—decisions, progress, architecture, glossaries, specs, and other custom data—connected as a project knowledge graph with explicit relationships. The system uses vector embeddings for semantic search to support Retrieval Augmented Generation (RAG) and more accurate information access. It supports multi-workspace workflows via workspace_id, with a primary STDIO deployment mode designed for tight IDE integration. The server provides a comprehensive set of MCP tools for interacting with the graph, including product context, active context, decisions, progress, system patterns, and custom data, plus facilities for linking items and exporting/importing data. Database migrations are managed via Alembic to ensure safe schema evolution. ConPort replaces file-based memory with a robust, queryable backend, and is designed to be IDE- and client-agnostic, usable from various MCP clients that support workspace-based operation. It includes setup guidance for local development and LLM strategy integrations.