# Hypermnesia MCP server

> Provides persistent, neuroscience-inspired memory for AI coding agents using local SQLite or PostgreSQL storage.

- Repository: https://github.com/cdeust/Cortex
- Page: https://hostedmcp.org/servers/cdeust/cortex
- Transport: stdio
- Stars: 77
- License: MIT
- Latest version: 4.24.0

## Install

### Claude Code

```sh
claude mcp add cortex -- uvx hypermnesia-mcp
```

### Cursor (`~/.cursor/mcp.json`)

```json
{
  "mcpServers": {
    "cortex": {
      "command": "uvx",
      "args": [
        "hypermnesia-mcp"
      ]
    }
  }
}
```

### VS Code (`.vscode/mcp.json`)

```json
{
  "servers": {
    "cortex": {
      "type": "stdio",
      "command": "uvx",
      "args": [
        "hypermnesia-mcp"
      ]
    }
  }
}
```

### Claude Desktop (`claude_desktop_config.json`)

```json
{
  "mcpServers": {
    "cortex": {
      "command": "uvx",
      "args": [
        "hypermnesia-mcp"
      ]
    }
  }
}
```

### Codex (`~/.codex/config.toml`)

```toml
[mcp_servers.cortex]
command = "uvx"
args = ["hypermnesia-mcp"]
```

### Gemini CLI (`~/.gemini/settings.json`)

```json
{
  "mcpServers": {
    "cortex": {
      "command": "uvx",
      "args": [
        "hypermnesia-mcp"
      ]
    }
  }
}
```

## FAQ

### Does this server require an internet connection or API key?

No, it runs locally by default with no account or API key required. Nothing leaves localhost unless you configure an integration.

### What databases can I use for storage?

It uses a local SQLite file by default, but you can also use PostgreSQL with pgvector if preferred.

### Can different AI agents share the same memory?

Yes, Claude Code and Codex can share Cortex memory by using the same configuration root and store.