Cortex
Persistent memory layer for AI coding tools (Claude, Cursor)
Start new thread

Cortex - Persistent memory layer for AI coding tools (Claude, Cursor)

by•
Cortex is a persistent memory layer for AI coding tools. Every new AI chat starts from zero — you re-explain your project, your stack, your decisions, every single time. Cortex fixes that. Connect any AI tool (Claude, Cursor, Copilot, Claude Code) to your GitHub repo through one remote MCP server, and it reads your actual codebase plus a living change-history file automatically. No copy-pasting context. No re-explaining. Your AI just remembers — across tools, across sessions, across months.

Add a comment

Replies

Best
Maker
📌
Every time I start a new AI session I have to re-explain everything. My stack, my project structure, what we decided last week, what failed. It's the same context dump, every single time. I built Cortex to fix this. Cortex is an MCP server that gives your AI tools persistent memory across sessions. You create a Space, connect your GitHub repo, and Cortex handles the rest. How it works: - You create a Space and describe your project once - Cortex creates a cortex.context file in your GitHub repo — the AI's permanent memory - Connect Claude Desktop or Cursor via MCP (one config paste) - AI automatically calls read_context() before every response — reads all your files, the context, and a change map - Every file the AI writes goes directly to GitHub as a real commit - cortex.map tracks every function/section change so the AI never contradicts its previous work - cortex.context evolves over time as the project grows No more re-explaining your stack. No more AI forgetting what you decided last session. No more contradictory changes. Works with: Claude Desktop, Cursor, VS Code Copilot, Windsurf, Claude Code CLI, Goose. It's free and open to try. Would love feedback from people who use AI heavily for coding.