An MCP server that provides AI assistants with long-term semantic memory using a Markdown-based knowledge base with hybrid search (vector + FTS5) and syntax-aware chunking. It supports idempotent ingestion, project isolation, and integrity linting.
