Skip to main content
MCP server CX Codex

AICortex — Claude Code MCP server for memory and integrations

The AICortex MCP server for Claude Code keeps an assistant’s memory, skills, and secrets on your own hardware. Connected models can load session context and reach configured services.

Apache-2.0 ★ 10 updated 2026-11-12

Open on GitHub

What AICortex does

AICortex stores persistent notes and skills, then gives an assistant user and project context at the start of a session. The same stored state can serve different models and devices. It runs on your own hardware, such as a NAS.

The server connects models to configured services and devices, including email and calendars, while keeping credentials in an encrypted vault. Other clients can use its REST API, and people manage settings through the /ui WebUI. Admin controls include users, services, and skills.

Scheduled jobs need a runner: options include a Claude Code cloud task, a ChatGPT plan, an API key, or local Ollama. Sign-in uses an OIDC provider. Credentials are sent only to the hosts they are bound to.

What it helps with

  • carry project notes across assistant sessions
  • connect an assistant to email and calendars
  • keep service credentials out of model context
  • run scheduled tasks through a chosen model

Example prompts for the agent

  • Remember how this project is organized and carry those notes into my next session.
  • Find a calendar skill and schedule a task for tomorrow.
  • Use the configured email connection to prepare a summary of new messages.

How to install AICortex

Install

The MCP registry has no launch command for it: see the repository README

Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more

Questions about AICortex

How do I install AICortex in Claude Code?

Run AICortex as an MCP server on your own hardware, such as a NAS. Once it is running, connect Claude Code to it over HTTPS.

What does AICortex do?

It stores assistant memory and skills, connects models to configured services and devices, and runs scheduled jobs. Settings are managed in the /ui WebUI.

What does AICortex need to run?

It needs hardware to host the server and an MCP-capable model. Sign-in uses an OIDC provider; scheduled jobs run through a Claude Code cloud task, a ChatGPT plan, an API key, or local Ollama.

Similar MCP servers

All MCP servers for Agent setup: memory, context, orchestration →

Project sponsors

Companies that keep this analytics open