# Claude vs Codex Statuslines

> Source: <https://jacob.gold/posts/claude-vs-codex-statuslines/>
> Published: 2026-06-28 21:32:06+00:00

# Claude vs Codex Statuslines

*
*

The Claude and Codex TUI coding agents both render a status line at the bottom of the interface. Configuring Claude’s status line works like configuring a shell prompt. The status line for Codex can only be configured to use a fixed set of built-in fields.

## Claude

Claude executes a [configured command](https://code.claude.com/docs/en/statusline) on each render, pipes its session state as JSON on stdin (model, cwd, and more), and renders whatever it writes to stdout. Mine prints the Git branch, model, token count, and number of running [claws](https://agentclaws.io), nodes, etc.

`~/.claude/settings.json`

```
{
  ...
  "statusLine": {
    "command": "clor statusline claude --template '{{with .ContextPercent}} {{.}}{{end}} {{.Model}} · 📁 {{.Dir}}{{with .Branch}} · 🌿 {{.}}{{end}}{{with .ClawSummary}} · 🦞 {{.}}{{end}}{{with .NodeSummary}} · 🖧  {{.}}{{end}}'",
    "type": "command"
  }
  ...
}
```

## Codex

Codex has a `/statusline`

command, but it only selects and reorders a fixed set of built-in fields. It can’t execute a command that prints arbitrary output, the way Claude can, so it can’t show a custom field like a claw count.

`~/.codex/config.toml`

```
...
[tui]
status_line = [
  "model-with-reasoning",
  "current-dir",
  "git-branch",
  "run-state",
  "permissions",
  "context-used",
  "context-window-size",
  "total-input-tokens",
  "total-output-tokens",
]
...
```

## The issue

People have been asking Codex for a script-backed status line in [issue #17827](https://github.com/openai/codex/issues/17827). It’s still open, with a long thread of plus-ones and no command-backed version shipped yet.

A customizable status line is a small thing on its own, but the ability to modify an agent for specific environments and purposes is one of the biggest opportunities to differentiate these tools.