Codex — what creators actually do with it

Updated 2026-10-22 · 17 posts · 8 S/A-tier

17 saved posts in this archive tag Codex, from 13 creators, posted between 2026-03 and 2026-10 (1 S-tier, 7 A-tier, 6 B-tier, 3 C-tier). Most sit in AI Saves, Vol. 8 (7), Creative (5), AI Saves, Vol. 7 (3). Most-saved creators: @nateherkai (3), @chase.h.ai (2), @tenfoldmarc (2). It most often appears alongside Claude Code (9), ChatGPT (4), Cursor (2).

Official site ↗

Editor's take

In these 17 posts Codex most often appears next to Claude Code (9 posts tag both), and usually as the second opinion. @jens.heitmann installs the Codex CLI plus a GitHub skill so Claude Code can call Codex for tests and audits, arguing the two models catch different things (audit setup). Other creators formalize the split: a Codex plugin inside Claude Code where one model plans and reviews while GPT-5.6 executes, stretching two subscriptions' limits (routing), and the Claudex skill, where Claude and GPT-6 Astra swap planner and auditor roles so neither grades its own work (Claudex). See the earned-autonomy router guide.

On its own, Codex is shown running operations as much as code. One setup gives it three markdown context files (an about-me interview, a banned-phrases style list, a company file) and three daily checkpoints for triage, delegation and an end-of-day audit (one-person company). Another runs six scheduled Codex tasks a day for a stock-trading experiment, all posting into one thread; there are no results yet, and the entry itself calls one-month outperformance noise (trading test). On the creative side, Codex drove Blender through a skills repo to build a 3D hero asset embedded with Three.js (2 posts tag Three.js), by a creator who had never used Blender (Blender build).

Cost is the recurring argument for it: one tier list ranks Codex alongside Claude Code and says it gives comparable output for less money (tier list), and another creator builds dashboards on top of an existing Codex subscription instead of paying for API calls (subscription UI). ChatGPT co-occurs in 4 posts and Cursor in 2. Several tags are incidental: news roundups and repo lists that name Codex next to Claude Code (Reverse Engineer Anything), and one prompt-pack post with no content at all (prompt pack).

Guides using it

Earned Autonomy — a three-tier router that demotes work to your own GPU3 of its postsBuild a three-tier task router (frontier, mid-tier, and a local Qwen 27B on one 24GB GPU) plus a zero tier for plain shell commands, wired into an existing orchestrator-worker setup. Each task class earns its way down the stack only…
Autonomous Agents in the Wild: What's Actually Working in 20262 of its postsSeparate working autonomous-agent patterns from demo-ware by building and stress-testing one narrow agent with real guardrails: a single task, one or two tools, fast feedback, hard budget and scope caps, persistent memory, and an…
Claude Mastery: From First Install to Advanced Agent Architecture2 of its postsTurn a bare Claude Code install into a mastered, measured setup for one real project: CLAUDE.md, skills, hooks, MCP/CLI tools, memory persistence, parallel worktrees, and cost controls - each adopted only after it beats the prior setup on…
Every App Is Headless Now — agents that build their own control surface2 of its postsMake one legacy tool that you or a client are licensed to use agent-drivable, lawfully, and decide honestly whether wrapping it is an arbitrage or a consulting treadmill.
Strategy Autoresearch — the judge is the product, not the agent2 of its postsSelf-rewriting loops (Karpathy-style autoresearch, bilevel coach loops, Microsoft SkillOpt) keep any edit that beats a benchmark. Point one at trading and it becomes an industrial overfitting machine — the backtest is the benchmark, and…
The Claude Harness Pattern: Context, Memory, and Cognitive Scaffolding2 of its postsProduce a working, version-controlled Claude harness for one real project: a CLAUDE.md of permanent rules, a primer.md that is rewritten every session, a skills/ folder of reusable workflows, and at least one hook that self-corrects…

All 17 posts · best first

SFour GitHub design repos — R3F, liquid glass, shader gradients, liquid chrome@cindie.zhu · 2026-09-02 · resource
AClaudex skill — Claude and GPT-6 Astra cross-audit plan and build@nateherkai · 2026-10-11 · skill
AReverse Engineer Anything — trending repo that explains how any app works@theericmichaud · 2026-10-11 · repo
AOpen-source AI week — OpenViking, ARTEMIS, GLM-5.3 FlashX, Unsloth Studio@drcintas · 2026-09-20 · news
ACodex + Blender — agent-built 3D hero asset embedded with Three.js@chase.h.ai · 2026-09-16 · technique
AGPT-6 Astra + Codex one-person company — context files and daily checkpoints@raycfu · 2026-09-09 · technique
AHerk Brain AI OS — local knowledge graph wired into a personal dashboard@nateherkai · 2026-09-07 · repo
ACodex + Claude Code — 3 systems to audit CC output with Codex@jens.heitmann · 2026-03-14 · technique
BSubscription token arbitrage — Claude Max vs Grok Ultra API-equivalent usage@duck.tate · 2026-09-12 · resource
BCursor 'Grok Bot' source-map leak — readable agent runtime rewired to any model@stridejosh.ai · 2026-08-25 · news
BFable 5 plus GPT-5.6 routing — Codex plugin inside Claude Code@cooper.simson · 2026-07-11 · technique
BAnthropic — for more details:@itsjustjlin · 2026-06-05 · technique
BAI tool tier list May 2026 — Codex and Claude Code top, chatbots C@chase.h.ai · 2026-05-15 · news
BSubscription-powered dashboards — custom UIs that drive Claude Code and Codex@rndyrbrts · 2026-04-30 · technique
CAI influencer pipeline — Claude-built persona monetized via Fanvue@tenfoldmarc · 2026-09-17 · technique
CGPT-6 Astra stock-trading challenge — scheduled Codex tasks in one thread@nateherkai · 2026-09-07 · demo
CGPT-6 Astra prompt pack — comment-bait with no content@tenfoldmarc · 2026-09-05 · resource

Pairs with

Count = posts that tag both tools.

Alternatives in this archive

Claude Code107 posts
Cursor9 posts
Gemini17 posts

Prompt systems that use it

All tool hubs Filter in the catalog All guides