Codex,
next to the rest of your team.
OpenCognit runs the Codex CLI in real terminals — side by side with Claude Code, Gemini, Kimi and Hermes. With your own login, several accounts, a check on every command and its tokens counted per project.
Codex and the others,
in one window.
The real Codex CLI,
with a control center around it.
Real Codex terminals
Every session is the real Codex CLI in its own terminal. Your settings and AGENTS.md work as before.
Work and private accounts
Several Codex logins side by side, each with its own config folder. When one hits its limit, OpenCognit shows it.
Checked before it runs
Codex starts with OpenCognit’s check hook. Dangerous commands are refused; ones that need approval wait for your OK in the app. Your ~/.codex stays untouched.
Local models too
Start “Codex local” against Ollama or LM Studio. With “Local models only”, the code never leaves your computer.
Tokens counted
The Usage page reads Codex’s local logs: tokens, API value and the share per project — nothing leaves your machine.
One project, many agents
Codex and Claude work on the same project side by side. The Neurons view shows which files each of them touches.
Three steps, no new account.
Install Codex
As usual, and log in.
Install OpenCognit
Linux (AppImage, .deb) or Windows. It finds Codex by itself.
Open a project
Click “+”, pick Codex — and start as many sessions as you need.
Questions about Codex
Do I need an OpenAI API key?
No. OpenCognit starts the Codex you already have installed, with your own login.
Does OpenCognit change my Codex settings?
No. The check hook is passed when Codex starts; your ~/.codex stays as it is.
Can Codex run with local models?
Yes. “Codex local” uses Ollama or LM Studio on your computer.
Does it run on macOS?
Not yet. OpenCognit is available for Linux and Windows.
What does it cost?
Free with up to 4 agents and 2 projects. Pro is a one-time purchase — no subscription on top of your plan.
