Connect the AIs you already use.
Cortex66 works with the AIs you have, not instead of them. You connect each one once, and you can take it back at any time.
Claude and ChatGPT on the web
In My AIs, choose Claude or ChatGPT and tap Connect. The AI's own consent screen opens; you approve there, and the Cortex66 connector is added to that AI. From then on it can read what you allow and take tasks from your chats.
The AI apps on your computer
Run npx cortex66, then cortex66 connect. The terminal shows a one-time code of 8 characters and opens cortex66.com, where you are already signed in. You approve the computer with one tap.
Then it looks for the AI apps it knows by their own config files: Claude Code, Cursor, VS Code, Codex CLI, Gemini CLI and Claude Desktop. For each app it asks first. Before it writes, it keeps a backup and changes only its own entry. --dry-run shows what it would change and changes nothing. ChatGPT desktop has no config file, so the Chrome extension is offered instead.
Check it and undo it
cortex66 status reads the files again and tells you, in plain words, what is connected. cortex66 disconnect removes exactly what it added and revokes the computer's token. You can also revoke the computer on the web; its next call is refused.
Web AIs through the extension
Some AIs, such as DeepSeek, have no connector. They can join through the Cortex66 Chrome extension, which types a task into the AI's own page after your approval and brings the answer back. This is being turned on step by step, so it may not show in your account yet.
You decide what each AI may read
Every connected AI has a trust level, shown under its name, for example "reads only this message". It is checked before anything leaves Cortex66, and you can change it.
See everything Cortex66 can do on Capabilities.