AI memory beyond the context window – Local Model Context Protocol server for AI coding assistants

Summary: Cortex is a local MCP server that extends AI coding assistants' memory by indexing committed code, documentation, and ADRs into vector embeddings for semantic retrieval. It integrates with Claude CLI and uses Postgres and Ollama for embeddings.

What it does

Cortex indexes project artifacts into vector embeddings to enable AI retrieval by meaning rather than keywords, avoiding duplicate implementations and repeated explanations.

Who it's for

It is designed for developers using AI coding assistants who need persistent, semantic project knowledge access beyond chat history.

Why it matters

It solves the limitation of AI context windows by providing memory that preserves approved project knowledge for efficient coding assistance.