Clarify that selecting "No" on first run also disables project skills

What feature would you like to see?

TL;DR

Selecting "No" on Codex’s first-run trust prompt marks the project as untrusted and silently disables loading skills from $CWD/.codex/skills, but this behavior is not documented.

Description

When running Codex for the first time, Codex shows the following prompt:

> You are running Codex in /path/to/project

  Since this folder is version controlled, you may wish to allow Codex to work in this folder without asking for approval.

› 1. Yes, allow Codex to work in this folder without asking for approval
  2. No, ask me to approve edits and commands

If option 2. No... is selected, Codex adds an untrusted entry to ~/.codex/config.toml.

The following document states that skills are loaded in order from $CWD/.codex/skills.
However, if the project is untrusted, they are not loaded.

https://developers.openai.com/codex/skills

This behavior is currently undocumented and non-obvious.
There is someone who is having the same problem. (#9665)

Request

Either of the following:

1. Clearly document that project-scoped skills are disabled when a project is marked as untrusted.
2. Warn users that selecting "No" on first run will also disable skill loading from the project directory.

I'd be happy to submit a PR for option 2 if that's the preferred direction.

Additional information

_No response_