# Run Codex in a sandbox
URL: https://docs.munito.ai/sandboxes/cloud-agents/codex

## Sign in [#sign-in]

Use one of these three methods.

### With an OpenAI API key [#with-an-openai-api-key]

Put the key in the `env` of the sandbox:

```json
{ "CODEX_API_KEY": "sk-..." }
```

- `codex exec` reads `CODEX_API_KEY` for each run.
- For the terminal UI, sign in with the key one time: `printenv CODEX_API_KEY | codex login --with-api-key`.
- OpenAI recommends an API key for automation. See [authentication](https://learn.chatgpt.com/docs/auth).

### With a ChatGPT plan [#with-a-chatgpt-plan]

Sign in over [SSH](https://docs.munito.ai/sandboxes/ports-and-ssh#ssh). Use a device code or your browser.

**With a device code:**

1. Turn on device code sign-in in the security settings of ChatGPT, or of your workspace.
2. Run `codex login --device-auth`.
3. Open the link on a device, and type the code.

**With your browser:**

1. Connect with `ssh -L 1455:localhost:1455`.
2. Run `codex login`.
3. Open the link in the browser of your computer. The forward sends the sign-in to the sandbox.

- Codex keeps the sign-in in `~/.codex/auth.json`. Each later `codex exec` uses it.
- Codex uses the usage limits of your plan.

### With the open models of Munito AI [#with-the-open-models-of-munito-ai]

1. Put your Munito AI key in the `env` of the sandbox as `MUNITO_API_KEY`.
2. Write this configuration to `~/.codex/config.toml`, with the files API or a command:

```toml
model = "zai-org/glm-5.3-flash"
model_provider = "munito"

[model_providers.munito]
name = "Munito"
base_url = "https://api.munito.ai/inference/v1"
env_key = "MUNITO_API_KEY"
wire_api = "responses"
```

[Codex with open models](https://docs.munito.ai/agents/codex) explains the provider, the model metadata and the regional hosts.

## Run a task [#run-a-task]

```bash
curl -sS --fail-with-body "https://api.munito.ai/sandboxes/v1/sandboxes/$SANDBOX/exec" \
  -H "Authorization: Bearer $MUNITO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"command": "cd /workspace && codex exec --skip-git-repo-check --dangerously-bypass-approvals-and-sandbox \"Write a Python script that prints the first 10 primes, and run it\"", "timeout": 1800}'
```

- `codex exec` runs one task, and then it stops. It writes the progress to standard error, and the last message to standard output.
- `--dangerously-bypass-approvals-and-sandbox` lets Codex run commands and change files with no question. The sandbox of Munito AI isolates Codex, so Codex does not need its own sandbox.
- Without this flag, `codex exec` can only read files.
- `--skip-git-repo-check` lets Codex work in a folder that is not a Git repository. In a cloned repository, remove it.
- `--json` gives each event of Codex as one line of JSON. With `"stream": true` in the request, the events come during the task.
- For a task of more than 30 minutes, start `codex exec` in the background with `nohup`, and read its log with later commands.

## Use the terminal UI [#use-the-terminal-ui]

1. Add your public key with [SSH](https://docs.munito.ai/sandboxes/ports-and-ssh#ssh).
2. Connect with the command that `GET /sandboxes/{sandbox_id}/ssh` gives.
3. Run `codex` in `/workspace`.

To keep the session after you disconnect, install `tmux` (`apt-get install -y tmux`), and start Codex in it.