Slash command `/fast status` is not handled by the CLI and is sent to the model as a plain text message

What version of Codex CLI is running?

codex-cli 0.153.2

What subscription do you have?

ChatGPT Pro 20x

Which model were you using?

gpt-6-astra

What platform is your computer?

Darwin 25.5.0 arm64 arm

What terminal emulator and version are you using (if applicable)?

iTerm2

Codex doctor report

What issue are you seeing?

When I type the slash command /fast status in the interactive TUI, the CLI does not
handle it as a command. Instead, the raw text /fast status is forwarded to the model
as a normal user message, so:

• no fast-mode status is printed by the CLI,
• no "unknown command" error is shown either,
• the model is left to guess what to do, reads local files (it opened a SKILL.md from
  a skill directory) and finally replies that it cannot read the session's fast-mode state.

In other words, the slash command silently degrades into prompt text, which both wastes
tokens and hides the fact that the command was not recognized/supported.

Screenshot

> /fast status

• I'll check the local instructions to confirm how to check the fast mode status.

• Explored
  └ Read SKILL.md (openai-docs skill)

• I can't directly read the Fast mode status of the current session; this `/fast status`
  was delivered to me as a normal message and no status result was returned.

What steps can reproduce the bug?

1. Start the interactive TUI: codex
2. Type the slash command /fast status and press Enter.
3. Observe that the CLI does not intercept the command: the text is submitted to the
   model as a regular user turn, the model starts reasoning/reading local files, and then
   replies that it cannot read the fast-mode status.

What is the expected behavior?

The official documentation explicitly documents /fast as a CLI command —
Speed → Fast mode
(zh-Hans: https://learn.chatgpt.com/zh-Hans/docs/agent-configuration/speed):

Use /fast on, /fast off, or /fast status in the CLI to change or inspect the
current setting. You can also persist the default with service_tier = "fast" plus
[features].fast_mode = true in config.toml. Fast mode is available in the ChatGPT
desktop app, Codex CLI, and IDE extension when you sign in with ChatGPT.

Expected, based on that documentation:

1. /fast status is handled locally by the CLI and never reaches the model.
   It should print the current setting, e.g.
   Fast mode: off (Standard) or Fast mode: on (1.5x speed, 2.5x credits),
   ideally including the effective model and where the setting comes from
   (session toggle vs. service_tier = "fast" / [features].fast_mode = true in config.toml).
2. No model turn, no tool calls, no token/credit consumption for a status query.
3. /fast on and /fast off behave symmetrically: they change the setting for the
   current session and echo the resulting state.
4. If Fast mode cannot apply in the current context (API key auth instead of ChatGPT
   sign-in, unsupported model/plan, feature disabled), the CLI should print a clear local
   error explaining why — still without forwarding the text to the model.
5. If this build genuinely does not implement /fast yet, then at minimum:
   • the CLI should fail locally with Unknown command: /fast plus a usage hint, and
   • /help and the documentation should state the minimum CLI version that supports it,
     so that documented behavior and shipped behavior stay consistent.

Actual: the command is treated as free-form prompt text and forwarded to the model.

Additional information

No response

Reproduce the issue from the interactive TUI with codex by submitting /fast status, then trace the TUI slash-command dispatch and the related config.toml settings. Confirm whether /fast is recognized in the shipped CLI and check /help for its documented availability. Done means the command is handled locally with a clear result or error, without creating a model turn or tool call.