HeyRocky

OpenAI Codex Provider Setup

What this page covers

This article explains OpenAI Codex Provider Setup as part of the Rocky/Hermes knowledge base. It is written for people building practical Hermes systems: terminal agents, Discord/Telegram gateways, scheduled jobs, website QA, voice workflows, and private operational assistants.

When to use it

Use this topic when you need to configure, verify, troubleshoot, or explain the Models & Providers layer of a Hermes setup.

Core idea

Hermes work should be handled as a verified workflow, not a guess. Start from the official Hermes documentation, make the smallest safe change, and finish with proof: command output, a test message, a generated artifact, a live page check, or a log entry.

Standard procedure

  • Read the relevant official documentation or built-in CLI help.
  • Identify whether this touches credentials, payments, private data, public pages, or destructive commands.
  • Run the smallest safe setup or inspection command.
  • Verify the result from the user's point of view.
  • Save durable procedures as a skill when the workflow is repeatable.

Verification checklist

  • The command, page, or integration was tested live.
  • No secrets or private customer data were exposed.
  • Public/private boundaries were checked if a website or gateway is involved.
  • The next action is clear and, if risky, requires human approval.

Related video

A matching Rocky YouTube script exists in the video script pack for this topic.

Official reference

Use the live Hermes docs as the source of truth: https://hermes-agent.nousresearch.com/docs

Related pages