{"slug": "codex-cli", "name": "Codex CLI", "category": "ai-coding", "category_name": "AI coding tools", "description": "OpenAI's open-source coding agent that runs in the terminal.", "recommenders": 47, "commenters": 124, "warners": 26, "score": 29.216, "rising": -7.42, "first": "2025-11-02", "latest": "2026-10-18", "why": "HN commenters use it daily as a main coding agent, report solid results on hands-on tasks like CAD and PCB design, and like that it is lighter on memory and battery than GUI apps.", "why_is_quote": false, "has_verdict": true, "sentiment": {"recommend": 47, "neutral": 54, "compare": 3, "warn": 28}, "spark": [6, 4, 7, 7, 6, 4, 2, 3, 2, 4, 1, 1], "listed": true, "loved": ["Real-task results", "Daily driver"], "caveat": "Background daemon", "short_quote": "Just get a ChatGPT $20/month  subscription and use the codex-cli.  I\u2019ve been using it every day since November and never run into limits", "kind": "", "icon": false, "domain": null, "github_repo": "github.com/openai/codex", "official_url": "https://github.com/openai/codex", "pricing": "open source", "aliases": ["Codex CLI", "OpenAI Codex CLI"], "verdict": [{"text": "HN commenters use it daily as a main coding agent", "cites": [{"id": 776931120, "item_id": 48558195, "url": "https://news.ycombinator.com/item?id=48558195", "by": "laurels-marts", "date": "2026-06-16", "version": null, "stance": "recommend", "quote": "still use Codex CLI as my main driver", "excerpt": "My company gave me cursor license after I had already been using Codex CLI for months and VS Code for a decade. I had absolutely no interest in their VS Code fork. The Agent Window was okay but buggy (eg wouldn\u2019t load branches on Ubuntu via WSL2). Overall used it a couple of times but still use Codex CLI as my main driver. Might try CC in the future esp. if they unban Fable.", "context": "", "use_case": "main coding agent", "switched_from": "", "story": "SpaceX Is Buying Cursor"}, {"id": 763838800, "item_id": 47739925, "url": "https://news.ycombinator.com/item?id=47739925", "by": "mrnoname", "date": "2026-04-12", "version": null, "stance": "recommend", "quote": "These days I reach for Codex CLI more than Claude Code.", "excerpt": "These days I reach for Codex CLI more than Claude Code. Not just for token economics, Codex has been catching up fast and is already ahead in some areas. It was open source from day one, arguably better code quality than the leaked Claude Code source, yet nobody cared. It's been improving quietly under the radar. I subscribe to ChatGPT Pro, but Ralph + Fast Mode can still burn through tokens. My company has a sizable", "context": "", "use_case": "", "switched_from": "", "story": "Show HN: Codex Workers AI Proxy \u2013 Use Cloudflare Workers AI models in Codex CLI"}, {"id": 771136050, "item_id": 48196003, "url": "https://news.ycombinator.com/item?id=48196003", "by": "m24927605", "date": "2026-05-19", "version": null, "stance": "recommend", "quote": "the AI coding agents I use daily \u2014 Claude Code, Codex CLI, Aider", "excerpt": "I built ait because the AI coding agents I use daily \u2014 Claude Code, Codex CLI, Aider \u2014 are fast individually, but no tool lets me put two of them in the same loop. You pick one. It runs. You hope it's right. That's not multi-agent \u2014 that's single-agent with chat history. ait is a local control plane that runs Claude Code, Codex, Aider, Gemini CLI, and Cursor as a team on the same task. One agent investigates, hands t", "context": "", "use_case": "AI coding", "switched_from": "", "story": "Show HN: Ait \u2013 Claude, Codex, and Aider as a team, on your laptop"}]}, {"text": ", report solid results on hands-on tasks like CAD and PCB design", "cites": [{"id": 801621841, "item_id": 50101365, "url": "https://news.ycombinator.com/item?id=50101365", "by": "edg5000", "date": "2026-10-15", "version": null, "stance": "recommend", "quote": "Highly recommend. Uses quite a lot of tokens.", "excerpt": "Astra in Codex CLI is good with FreeCAD if you let it use the FreeCAD Python runtime in a headless session. It is good at relying purely on natural language to design and revise. To follow specific CAD techniques you need additional prompting. Highly recommend. Uses quite a lot of tokens.", "context": "", "use_case": "driving FreeCAD with AI", "switched_from": "", "story": "Ask HN: CAD for wood working with LLM support?"}, {"id": 773903968, "item_id": 48368998, "url": "https://news.ycombinator.com/item?id=48368998", "by": "mapontosevenths", "date": "2026-06-02", "version": null, "stance": "recommend", "quote": "I literally did this yesterday with solid results using Codex CLI", "excerpt": "> I could only get about 80% of what I wanted together with my hacky workflow. I literally did this yesterday with solid results using Codex CLI. I used xhigh thinking and gpt 5.5. I had it use KiCad directly via cli rather than via MCP, and I did make Claude Opus review it's work after every round. I got what I think will be a working revision A in about 10 hours of tinkering spread over a few days.", "context": "", "use_case": "PCB design via KiCad CLI", "switched_from": "", "story": "Adafruit Receives Demand Letter from Fenwick Legal Counsel on Behalf of Flux.ai"}, {"id": 763615360, "item_id": 47725960, "url": "https://news.ycombinator.com/item?id=47725960", "by": "osti", "date": "2026-04-10", "version": null, "stance": "recommend", "quote": "I was using codex cli with 5.4xhigh.", "excerpt": "I was using codex cli with 5.4xhigh. So it was able to iteratively improve from simple prompts on my part (can you give some architectural ideas to improve the performance? And once it does, I just say can you implement and benchmark it). I think it was a bit like Karpathy's autoresearch, except I was doing manual promoting... Though I feel I could definitely be removed from that equation.", "context": "", "use_case": "", "switched_from": "", "story": "ChatGPT Pro now starts at $100/month"}]}, {"text": ", and like that it is lighter on memory and battery than GUI apps.", "cites": [{"id": 788536161, "item_id": 49283510, "url": "https://news.ycombinator.com/item?id=49283510", "by": "meerita", "date": "2026-08-13", "version": null, "stance": "recommend", "quote": "Both are really good, they don't eat memory like the GUI apps.", "excerpt": "I use Claude Code CLI, Codex Cli. Both are really good, they don't eat memory like the GUI apps.", "context": "", "use_case": "Terminal coding agent", "switched_from": "", "story": "ChatGPT Desktop (Codex Desktop) for Linux"}, {"id": 764855361, "item_id": 47803460, "url": "https://news.ycombinator.com/item?id=47803460", "by": "JodieBenitez", "date": "2026-04-17", "version": null, "stance": "recommend", "quote": "I use Codex CLI now, it doesn't drain the battery.", "excerpt": "Ditched it for this very reason... it used to be fine before. I use Codex CLI now, it doesn't drain the battery. I prefer the desktop app but the CLI is ok.", "context": "", "use_case": "Coding agent CLI", "switched_from": "Codex", "story": "Codex for Almost Everything"}]}, {"text": " The most common complaint is a background auto-update daemon added in recent versions", "cites": [{"id": 800987760, "item_id": 50061735, "url": "https://news.ycombinator.com/item?id=50061735", "by": "vunderba", "date": "2026-10-12", "version": null, "stance": "warn", "quote": "they installed this persistent nonsense", "excerpt": "Yep, the real issue is that it was something they didn\u2019t have before, and then with Codex 0.157.0 enabled, they installed this persistent nonsense [1] that survives after the last instance of Codex has exited on your machine, which is just gross. codex features disable daemon_auto_start codex app-server daemon stop pkill -f 'app-server daemon pid-update-loop' [1] - https://github.com/openai/codex/issues/48195", "context": "", "use_case": "", "switched_from": "", "story": "Codex CLI installs a persistent, self-updating background daemon by default"}, {"id": 800988032, "item_id": 50061752, "url": "https://news.ycombinator.com/item?id=50061752", "by": "dllu", "date": "2026-10-12", "version": null, "stance": "warn", "quote": "it is extremely annoying how many bugs Codex has. I basically can't trust it to execute on any goal", "excerpt": "The daemon has many bugs. For example, when auto-updating, it would randomly ruin the permissions and block the goal as .git became read-only. Generally, it is extremely annoying how many bugs Codex has. I basically can't trust it to execute on any goal since it would often get blocked every few hours due to random bugs. https://github.com/openai/codex/issues/52251#issuecomment-6090672165", "context": "", "use_case": "executing coding goals", "switched_from": "", "story": "Codex CLI installs a persistent, self-updating background daemon by default"}, {"id": 800990256, "item_id": 50061891, "url": "https://news.ycombinator.com/item?id=50061891", "by": "ilitirit", "date": "2026-10-12", "version": null, "stance": "warn", "quote": "I don't know how and why they ruined this CLI with needless \"features\".", "excerpt": "I use it at work. On Windows it also launches a new console window every it executes (?) a command. You should definitely disabled this using --no-daemon. I don't know how and why they ruined this CLI with needless \"features\". It was perfectly usable two weeks ago.", "context": "uses it at work on Windows", "use_case": "AI coding at work on Windows", "switched_from": "", "story": "Codex CLI installs a persistent, self-updating background daemon by default"}, {"id": 801001104, "item_id": 50062569, "url": "https://news.ycombinator.com/item?id=50062569", "by": "AndreasBackx", "date": "2026-10-12", "version": null, "stance": "warn", "quote": "It's managing the installation by itself not using existing package managers.", "excerpt": "I feel like so many of the new AI tooling is reinventing the wheel. We've figured out how to update software via package managers. Do we need a daemon to continually check for updates and do we need shell scripts to circumvent package managers? Have a look at https://chatgpt.com/codex/install.sh. It's managing the installation by itself not using existing package managers. There's a good point to making it easy to in", "context": "", "use_case": "", "switched_from": "", "story": "Codex CLI installs a persistent, self-updating background daemon by default"}]}, {"text": ", along with bugs that block long-running goals.", "cites": [{"id": 801001888, "item_id": 50062618, "url": "https://news.ycombinator.com/item?id=50062618", "by": "dllu", "date": "2026-10-12", "version": null, "stance": "warn", "quote": "goal-breaking bugs I've run into", "excerpt": "My complaint applies specifically to the CLI. I have never used the gui app. Here are the various goal-breaking bugs I've run into: - Goals get blocked due to .git and other stuff becoming randomly read-only when daemon auto-updates in the background (the linked issue in my comment above) - Goals get blocked due to transient \"Selected model is at capacity\" - Goals get blocked due to \"Codex ran out of room in the mode", "context": "", "use_case": "coding agent CLI", "switched_from": "", "story": "Codex CLI installs a persistent, self-updating background daemon by default"}, {"id": 800988032, "item_id": 50061752, "url": "https://news.ycombinator.com/item?id=50061752", "by": "dllu", "date": "2026-10-12", "version": null, "stance": "warn", "quote": "it is extremely annoying how many bugs Codex has. I basically can't trust it to execute on any goal", "excerpt": "The daemon has many bugs. For example, when auto-updating, it would randomly ruin the permissions and block the goal as .git became read-only. Generally, it is extremely annoying how many bugs Codex has. I basically can't trust it to execute on any goal since it would often get blocked every few hours due to random bugs. https://github.com/openai/codex/issues/52251#issuecomment-6090672165", "context": "", "use_case": "executing coding goals", "switched_from": "", "story": "Codex CLI installs a persistent, self-updating background daemon by default"}]}], "praise": [{"label": "Real-task results", "summary": "Commenters report solid results on CAD, PCB design and performance work.", "quotes": [{"id": 773903968, "item_id": 48368998, "url": "https://news.ycombinator.com/item?id=48368998", "by": "mapontosevenths", "date": "2026-06-02", "version": null, "stance": "recommend", "quote": "I literally did this yesterday with solid results using Codex CLI", "excerpt": "> I could only get about 80% of what I wanted together with my hacky workflow. I literally did this yesterday with solid results using Codex CLI. I used xhigh thinking and gpt 5.5. I had it use KiCad directly via cli rather than via MCP, and I did make Claude Opus review it's work after every round. I got what I think will be a working revision A in about 10 hours of tinkering spread over a few days.", "context": "", "use_case": "PCB design via KiCad CLI", "switched_from": "", "story": "Adafruit Receives Demand Letter from Fenwick Legal Counsel on Behalf of Flux.ai"}, {"id": 801621841, "item_id": 50101365, "url": "https://news.ycombinator.com/item?id=50101365", "by": "edg5000", "date": "2026-10-15", "version": null, "stance": "recommend", "quote": "Highly recommend. Uses quite a lot of tokens.", "excerpt": "Astra in Codex CLI is good with FreeCAD if you let it use the FreeCAD Python runtime in a headless session. It is good at relying purely on natural language to design and revise. To follow specific CAD techniques you need additional prompting. Highly recommend. Uses quite a lot of tokens.", "context": "", "use_case": "driving FreeCAD with AI", "switched_from": "", "story": "Ask HN: CAD for wood working with LLM support?"}, {"id": 763615360, "item_id": 47725960, "url": "https://news.ycombinator.com/item?id=47725960", "by": "osti", "date": "2026-04-10", "version": null, "stance": "recommend", "quote": "I was using codex cli with 5.4xhigh.", "excerpt": "I was using codex cli with 5.4xhigh. So it was able to iteratively improve from simple prompts on my part (can you give some architectural ideas to improve the performance? And once it does, I just say can you implement and benchmark it). I think it was a bit like Karpathy's autoresearch, except I was doing manual promoting... Though I feel I could definitely be removed from that equation.", "context": "", "use_case": "", "switched_from": "", "story": "ChatGPT Pro now starts at $100/month"}]}, {"label": "Daily driver", "summary": "Several use it as their main agent, some preferring it over Claude Code or Cursor.", "quotes": [{"id": 776931120, "item_id": 48558195, "url": "https://news.ycombinator.com/item?id=48558195", "by": "laurels-marts", "date": "2026-06-16", "version": null, "stance": "recommend", "quote": "still use Codex CLI as my main driver", "excerpt": "My company gave me cursor license after I had already been using Codex CLI for months and VS Code for a decade. I had absolutely no interest in their VS Code fork. The Agent Window was okay but buggy (eg wouldn\u2019t load branches on Ubuntu via WSL2). Overall used it a couple of times but still use Codex CLI as my main driver. Might try CC in the future esp. if they unban Fable.", "context": "", "use_case": "main coding agent", "switched_from": "", "story": "SpaceX Is Buying Cursor"}, {"id": 763838800, "item_id": 47739925, "url": "https://news.ycombinator.com/item?id=47739925", "by": "mrnoname", "date": "2026-04-12", "version": null, "stance": "recommend", "quote": "These days I reach for Codex CLI more than Claude Code.", "excerpt": "These days I reach for Codex CLI more than Claude Code. Not just for token economics, Codex has been catching up fast and is already ahead in some areas. It was open source from day one, arguably better code quality than the leaked Claude Code source, yet nobody cared. It's been improving quietly under the radar. I subscribe to ChatGPT Pro, but Ralph + Fast Mode can still burn through tokens. My company has a sizable", "context": "", "use_case": "", "switched_from": "", "story": "Show HN: Codex Workers AI Proxy \u2013 Use Cloudflare Workers AI models in Codex CLI"}, {"id": 788698752, "item_id": 49293672, "url": "https://news.ycombinator.com/item?id=49293672", "by": "karanraina", "date": "2026-08-13", "version": null, "stance": "recommend", "quote": "Codex is both faster and better.", "excerpt": "I use codex cli. It doens't have a UI for linux. I use opencode for that but it doens't work that well with the new 5.6 model suite from openai. Codex is both faster and better. I would have loved if there was a way to use a GUI that worked like codex. Gave this a try. I think this might be it. Please consider atleast adding MCP support if you can. That'll help", "context": "Linux user", "use_case": "coding agent", "switched_from": "", "story": "Launch HN: Bullet (YC S26) \u2013 A Faster Coding Agent"}]}, {"label": "Flexible setup", "summary": "Runs in containers, over SSH on a VM, or with OpenRouter models.", "quotes": [{"id": 769689921, "item_id": 48105620, "url": "https://news.ycombinator.com/item?id=48105620", "by": "nkmnz", "date": "2026-05-12", "version": null, "stance": "recommend", "quote": "For codex, I run the cli in my working dir prefixed with `CODEX_HOME=./codex-home codex`", "excerpt": "It's possible to use subscriptions! I run them in containers. For claude, I use `claude setup-token`, put the token into a local auth.json and mount that. For codex, I run the cli in my working dir prefixed with `CODEX_HOME=./codex-home codex` and mount that whole `codex-home` directory - done :)", "context": "", "use_case": "running in containers on subscription", "switched_from": "", "story": null}, {"id": 788537792, "item_id": 49283612, "url": "https://news.ycombinator.com/item?id=49283612", "by": "preek", "date": "2026-08-13", "version": null, "stance": "recommend", "quote": "I have a codex cli instance on a Debian VM connected to ChatGPT on iOS via SSH. That works since.. about a month and has pretty good UX.", "excerpt": "I have a codex cli instance on a Debian VM connected to ChatGPT on iOS via SSH. That works since.. about a month and has pretty good UX. Before, I used the shellfish terminal and tmux which also worked really good, but to be honest (even though I'm a big terminal fan), ChatGPT on iOS sometimes has a smoother UX than a terminal.", "context": "terminal fan", "use_case": "remote coding from phone", "switched_from": "", "story": "ChatGPT Desktop (Codex Desktop) for Linux"}, {"id": 778262850, "item_id": 48641428, "url": "https://news.ycombinator.com/item?id=48641428", "by": "PinguTS", "date": "2026-06-23", "version": null, "stance": "recommend", "quote": "CodeX with OpenRouter", "excerpt": "Why not using some other coding IDE with OpenRouter? There you can also use the models. You can even switch models easily depending on your needs and spent. There are OpenSource versions of CodeX. For example: * OpenCode: https://medium.com/codex/kickstart-opencode-with-openrouter-afb2b6e284e5 * CodeX with OpenRouter: https://openrouter.ai/docs/cookbook/coding-agents/codex-cli", "context": "", "use_case": "", "switched_from": "", "story": "Ask HN: Anthropic banned me from using Claude Code and I don't know what to do"}]}, {"label": "Lightweight", "summary": "Uses less memory than GUI apps and doesn't drain the battery.", "quotes": [{"id": 788536161, "item_id": 49283510, "url": "https://news.ycombinator.com/item?id=49283510", "by": "meerita", "date": "2026-08-13", "version": null, "stance": "recommend", "quote": "Both are really good, they don't eat memory like the GUI apps.", "excerpt": "I use Claude Code CLI, Codex Cli. Both are really good, they don't eat memory like the GUI apps.", "context": "", "use_case": "Terminal coding agent", "switched_from": "", "story": "ChatGPT Desktop (Codex Desktop) for Linux"}, {"id": 764855361, "item_id": 47803460, "url": "https://news.ycombinator.com/item?id=47803460", "by": "JodieBenitez", "date": "2026-04-17", "version": null, "stance": "recommend", "quote": "I use Codex CLI now, it doesn't drain the battery.", "excerpt": "Ditched it for this very reason... it used to be fine before. I use Codex CLI now, it doesn't drain the battery. I prefer the desktop app but the CLI is ok.", "context": "", "use_case": "Coding agent CLI", "switched_from": "Codex", "story": "Codex for Almost Everything"}]}, {"label": "Open source", "summary": "Commenters note it has been open source from the start.", "quotes": [{"id": 763838800, "item_id": 47739925, "url": "https://news.ycombinator.com/item?id=47739925", "by": "mrnoname", "date": "2026-04-12", "version": null, "stance": "recommend", "quote": "These days I reach for Codex CLI more than Claude Code.", "excerpt": "These days I reach for Codex CLI more than Claude Code. Not just for token economics, Codex has been catching up fast and is already ahead in some areas. It was open source from day one, arguably better code quality than the leaked Claude Code source, yet nobody cared. It's been improving quietly under the radar. I subscribe to ChatGPT Pro, but Ralph + Fast Mode can still burn through tokens. My company has a sizable", "context": "", "use_case": "", "switched_from": "", "story": "Show HN: Codex Workers AI Proxy \u2013 Use Cloudflare Workers AI models in Codex CLI"}, {"id": 790110096, "item_id": 49381881, "url": "https://news.ycombinator.com/item?id=49381881", "by": "amluto", "date": "2026-08-20", "version": null, "stance": "recommend", "quote": "Looks like it's still a more-or-less-open-source product called Codex CLI.", "excerpt": "Since when? https://learn.chatgpt.com/docs/codex/cli https://github.com/openai/codex Looks like it's still a more-or-less-open-source product called Codex CLI. FWIW, there have been issues with various versions of ChatGPT not knowing what Codex is -- at least in the (IIRC) GPT-5.3 timeframe, there were serious issues with the knowledge cutoff.", "context": "", "use_case": "", "switched_from": "", "story": "Asana cleared 5 years of engineering work in 2 weeks with Codex"}]}], "caveats": [{"label": "Background daemon", "summary": "A persistent auto-update daemon outlives sessions and causes problems, prompting users to disable it.", "quotes": [{"id": 800987760, "item_id": 50061735, "url": "https://news.ycombinator.com/item?id=50061735", "by": "vunderba", "date": "2026-10-12", "version": null, "stance": "warn", "quote": "they installed this persistent nonsense", "excerpt": "Yep, the real issue is that it was something they didn\u2019t have before, and then with Codex 0.157.0 enabled, they installed this persistent nonsense [1] that survives after the last instance of Codex has exited on your machine, which is just gross. codex features disable daemon_auto_start codex app-server daemon stop pkill -f 'app-server daemon pid-update-loop' [1] - https://github.com/openai/codex/issues/48195", "context": "", "use_case": "", "switched_from": "", "story": "Codex CLI installs a persistent, self-updating background daemon by default"}, {"id": 800988032, "item_id": 50061752, "url": "https://news.ycombinator.com/item?id=50061752", "by": "dllu", "date": "2026-10-12", "version": null, "stance": "warn", "quote": "it is extremely annoying how many bugs Codex has. I basically can't trust it to execute on any goal", "excerpt": "The daemon has many bugs. For example, when auto-updating, it would randomly ruin the permissions and block the goal as .git became read-only. Generally, it is extremely annoying how many bugs Codex has. I basically can't trust it to execute on any goal since it would often get blocked every few hours due to random bugs. https://github.com/openai/codex/issues/52251#issuecomment-6090672165", "context": "", "use_case": "executing coding goals", "switched_from": "", "story": "Codex CLI installs a persistent, self-updating background daemon by default"}, {"id": 801001104, "item_id": 50062569, "url": "https://news.ycombinator.com/item?id=50062569", "by": "AndreasBackx", "date": "2026-10-12", "version": null, "stance": "warn", "quote": "It's managing the installation by itself not using existing package managers.", "excerpt": "I feel like so many of the new AI tooling is reinventing the wheel. We've figured out how to update software via package managers. Do we need a daemon to continually check for updates and do we need shell scripts to circumvent package managers? Have a look at https://chatgpt.com/codex/install.sh. It's managing the installation by itself not using existing package managers. There's a good point to making it easy to in", "context": "", "use_case": "", "switched_from": "", "story": "Codex CLI installs a persistent, self-updating background daemon by default"}]}, {"label": "Frequent bugs", "summary": "Bugs like broken mouse copy and failed context compaction block work.", "quotes": [{"id": 800988032, "item_id": 50061752, "url": "https://news.ycombinator.com/item?id=50061752", "by": "dllu", "date": "2026-10-12", "version": null, "stance": "warn", "quote": "it is extremely annoying how many bugs Codex has. I basically can't trust it to execute on any goal", "excerpt": "The daemon has many bugs. For example, when auto-updating, it would randomly ruin the permissions and block the goal as .git became read-only. Generally, it is extremely annoying how many bugs Codex has. I basically can't trust it to execute on any goal since it would often get blocked every few hours due to random bugs. https://github.com/openai/codex/issues/52251#issuecomment-6090672165", "context": "", "use_case": "executing coding goals", "switched_from": "", "story": "Codex CLI installs a persistent, self-updating background daemon by default"}, {"id": 801080256, "item_id": 50067516, "url": "https://news.ycombinator.com/item?id=50067516", "by": "srigi", "date": "2026-10-13", "version": null, "stance": "warn", "quote": "Recently they broke the most essential thing you can do in TUI", "excerpt": "Recently they broke the most essential thing you can do in TUI - copying the text selected with the mouse: https://github.com/openai/codex/issues/48125", "context": "", "use_case": "", "switched_from": "", "story": "Codex CLI installs a persistent, self-updating background daemon by default"}, {"id": 795884400, "item_id": 49742775, "url": "https://news.ycombinator.com/item?id=49742775", "by": "nowittyusername", "date": "2026-09-17", "version": null, "stance": "warn", "quote": "codex cli compaction is garbage so I made my own custom implementation", "excerpt": "I been working and making things with AI agents since Windsurf days, here's what works for me after experimenting and working for a while with these things. As far as the agent goes I find whatever the latest OpenAI model is out worked best for me. This company has burned me the least and I like the models. I tried many but consistency of OpenAI models cant be beat IMO. Though we are post honeymoon phase now I feel l", "context": "", "use_case": "coding agent", "switched_from": "", "story": "Show HN: Share your AI Setup, Learn from others"}]}, {"label": "Windows issues", "summary": "Windows users report screen flashing and new console windows per command.", "quotes": [{"id": 798455424, "item_id": 49903464, "url": "https://news.ycombinator.com/item?id=49903464", "by": "coverband", "date": "2026-09-30", "version": null, "stance": "warn", "quote": "most Windows users had to rollback/downgrade to an older version because of the annoying excessive terminal screen flashing issue", "excerpt": "I love the fact that they're advertising the latest Codex Cli while most Windows users had to rollback/downgrade to an older version because of the annoying excessive terminal screen flashing issue (https://github.com/openai/codex/issues/48074) (I see a newer version was posted an hour ago. Let's see if the issue is now fixed...)", "context": "Windows user", "use_case": "", "switched_from": "", "story": "DevDay 2026 Recap"}, {"id": 800990256, "item_id": 50061891, "url": "https://news.ycombinator.com/item?id=50061891", "by": "ilitirit", "date": "2026-10-12", "version": null, "stance": "warn", "quote": "I don't know how and why they ruined this CLI with needless \"features\".", "excerpt": "I use it at work. On Windows it also launches a new console window every it executes (?) a command. You should definitely disabled this using --no-daemon. I don't know how and why they ruined this CLI with needless \"features\". It was perfectly usable two weeks ago.", "context": "uses it at work on Windows", "use_case": "AI coding at work on Windows", "switched_from": "", "story": "Codex CLI installs a persistent, self-updating background daemon by default"}]}, {"label": "Permissions", "summary": "Users report it acting outside granted permissions and call its permission model broken.", "quotes": [{"id": 801073680, "item_id": 50067105, "url": "https://news.ycombinator.com/item?id=50067105", "by": "crossroadsguy", "date": "2026-10-13", "version": null, "stance": "warn", "quote": "one day Codex+GLM went ahead and did things in my ~/.dotfiles folder which I had granted only READ access for", "excerpt": "I was delaying it but one day Codex+GLM went ahead and did things in my ~/.dotfiles folder which I had granted only READ access for. I asked and I was given a reason which I didn't understand and didn't really want to. Right then I used sandbox-exec/safehouse (macOS) and tightened every harness installed (and the one not even installed, lest I install it later and forget to add/track them in the sb zsh init script). ", "context": "macOS user", "use_case": "AI coding agent", "switched_from": "", "story": "Codex CLI installs a persistent, self-updating background daemon by default"}, {"id": 796504885, "item_id": 49781555, "url": "https://news.ycombinator.com/item?id=49781555", "by": "srcreigh", "date": "2026-09-21", "version": null, "stance": "warn", "quote": "Codex CLI permissions for example are extremely broken", "excerpt": "So the agent-substrate checks a _ton_ of boxes. Almost all of the things it offers should be table stakes for everywhere we run not only agents but most software. https://github.com/agent-substrate/substrate (For context I built something very similar to this the past 2 weeks for my homelab, trying to solve many of these problems. This comment is an edited version of an unreleased blog post I wrote last week.) - Run ", "context": "", "use_case": "", "switched_from": "", "story": "AX \u2013 Google\u2019s Open Agentic Orchestrator"}]}, {"label": "Token usage", "summary": "It can burn through a lot of tokens.", "quotes": [{"id": 801621841, "item_id": 50101365, "url": "https://news.ycombinator.com/item?id=50101365", "by": "edg5000", "date": "2026-10-15", "version": null, "stance": "recommend", "quote": "Highly recommend. Uses quite a lot of tokens.", "excerpt": "Astra in Codex CLI is good with FreeCAD if you let it use the FreeCAD Python runtime in a headless session. It is good at relying purely on natural language to design and revise. To follow specific CAD techniques you need additional prompting. Highly recommend. Uses quite a lot of tokens.", "context": "", "use_case": "driving FreeCAD with AI", "switched_from": "", "story": "Ask HN: CAD for wood working with LLM support?"}, {"id": 763838800, "item_id": 47739925, "url": "https://news.ycombinator.com/item?id=47739925", "by": "mrnoname", "date": "2026-04-12", "version": null, "stance": "recommend", "quote": "These days I reach for Codex CLI more than Claude Code.", "excerpt": "These days I reach for Codex CLI more than Claude Code. Not just for token economics, Codex has been catching up fast and is already ahead in some areas. It was open source from day one, arguably better code quality than the leaked Claude Code source, yet nobody cared. It's been improving quietly under the radar. I subscribe to ChatGPT Pro, but Ralph + Fast Mode can still burn through tokens. My company has a sizable", "context": "", "use_case": "", "switched_from": "", "story": "Show HN: Codex Workers AI Proxy \u2013 Use Cloudflare Workers AI models in Codex CLI"}]}, {"label": "Harness quality", "summary": "Some find the harness bare bones or its system prompt limiting.", "quotes": [{"id": 790478413, "item_id": 49404900, "url": "https://news.ycombinator.com/item?id=49404900", "by": "c0rruptbytes", "date": "2026-08-22", "version": null, "stance": "warn", "quote": "Codex CLI is pretty bare bones in a bad way", "excerpt": "Pi by itself is more than capable, OMP is okay but you really don't need much for a great harness (these models are RL trained to hell to be a coding agent, sometimes less is more) I run a lot of SlopCodeBench - https://github.com/michaelasper/benchmarks Fable/Sol/GLM 5.3/Kimi are its league (in that order) Deepseek/Opus is solid Qwen 27B is the floor - there's no reason to use Sonnet/Terra/Haiku For everyday activit", "context": "", "use_case": "", "switched_from": "", "story": "A week of using Codex more than Claude"}, {"id": 794997649, "item_id": 49687353, "url": "https://news.ycombinator.com/item?id=49687353", "by": "aleda145", "date": "2026-09-13", "version": null, "stance": "warn", "quote": "it feels like its extensive system prompt cripples the agent a lot vs Kimi 2.7", "excerpt": "I open sourced kavla, the SQL canvas I'm working on: https://github.com/aleda145/kavla README is very bare bones, needs a video or a picture too. The website should tell you some more things: https://kavla.dev (and now the demo does not require a login (also need to remake the demo to fit the new self hosted version)) It's still lacking an analytics agent for the local version though. I've explored using codex CLI, b", "context": "", "use_case": "analytics agent", "switched_from": "", "story": "Ask HN: What are you working on? (September 2026)"}]}], "who": [{"by": "entropie", "context": "NixOS user", "url": "https://news.ycombinator.com/item?id=50062074", "date": "2026-10-12"}, {"by": "alex-moon", "context": "supervises the agent closely", "url": "https://news.ycombinator.com/item?id=49651374", "date": "2026-09-10"}, {"by": "karanraina", "context": "Linux user", "url": "https://news.ycombinator.com/item?id=49293672", "date": "2026-08-13"}, {"by": "preek", "context": "terminal fan", "url": "https://news.ycombinator.com/item?id=49283612", "date": "2026-08-13"}, {"by": "danielnexaaly", "context": "AI-assisted technical operator", "url": "https://news.ycombinator.com/item?id=49251761", "date": "2026-08-10"}, {"by": "nsingh2", "context": "former heavy Cursor user", "url": "https://news.ycombinator.com/item?id=47455950", "date": "2026-03-20"}, {"by": "raw_anon_1111", "context": "Personal $20/month ChatGPT plan; $750/month corporate Claude allowance", "url": "https://news.ycombinator.com/item?id=47175772", "date": "2026-02-26"}, {"by": "leonflexo", "context": "learning VPS and NGINX", "url": "https://news.ycombinator.com/item?id=46372473", "date": "2025-12-23"}, {"by": "pqs", "context": "exported 25k Evernote notes", "url": "https://news.ycombinator.com/item?id=46300683", "date": "2025-12-17"}, {"by": "dloss", "context": "hobby programmer for 30+ years", "url": "https://news.ycombinator.com/item?id=46181970", "date": "2025-12-07"}, {"by": "not-a-cat", "context": "Founded Tufalabs about a year ago", "url": "https://news.ycombinator.com/item?id=45790656", "date": "2025-11-02"}], "tags": ["for terminal users", "for Linux users", "for supervised coding", "for multi-agent workflows", "for containerized setups"], "use_cases": ["ai coding", "coding agent", "coding agent cli", "ai coding agent", "terminal coding agent", "daily coding"], "compared": [{"slug": "claude-code", "name": "Claude Code", "co": 67, "switched_from": 3, "switched_to": 0}, {"slug": "opencode", "name": "OpenCode", "co": 14, "switched_from": 0, "switched_to": 0}, {"slug": "gpt-5", "name": "GPT-5", "co": 14, "switched_from": 0, "switched_to": 0}, {"slug": "claude", "name": "Claude", "co": 10, "switched_from": 1, "switched_to": 0}, {"slug": "chatgpt", "name": "ChatGPT", "co": 11, "switched_from": 0, "switched_to": 0}, {"slug": "claude-opus", "name": "Claude Opus", "co": 11, "switched_from": 0, "switched_to": 0}, {"slug": "pi", "name": "Pi", "co": 7, "switched_from": 0, "switched_to": 2}, {"slug": "gemini-cli", "name": "Gemini CLI", "co": 11, "switched_from": 0, "switched_to": 0}], "makers": [{"id": 766228256, "item_id": 47889266, "url": "https://news.ycombinator.com/item?id=47889266", "by": "kordlessagain", "date": "2026-04-24", "version": null, "stance": "neutral", "quote": "If anyone wants Codex CLI containers with various MCP tools available, I built this", "excerpt": "If anyone wants Codex CLI containers with various MCP tools available, I built this: https://deepbluedynamics.com/nemesis", "context": "", "use_case": "", "switched_from": "", "story": "Introducing GPT-5.5"}], "threads": [{"url": "https://news.ycombinator.com/item?id=50060694", "title": "Codex CLI installs a persistent, self-updating background daemon by default", "mentions": 9}, {"url": "https://news.ycombinator.com/item?id=46737630", "title": "Unrolling the Codex Agent Loop", "mentions": 6}, {"url": "https://news.ycombinator.com/item?id=46250332", "title": "OpenAI are quietly adopting skills, now available in ChatGPT and Codex CLI", "mentions": 6}, {"url": "https://news.ycombinator.com/item?id=45982649", "title": "Building more with GPT-5.1-Codex-Max", "mentions": 4}, {"url": "https://news.ycombinator.com/item?id=49281916", "title": "ChatGPT Desktop (Codex Desktop) for Linux", "mentions": 3}, {"url": "https://news.ycombinator.com/item?id=47311484", "title": "JSLinux Now Supports x86_64", "mentions": 3}], "sentiment_by_month": {"2025-11": {"recommend": 6, "neutral": 3, "warn": 5}, "2025-12": {"neutral": 7, "compare": 1, "warn": 1, "recommend": 4}, "2026-01": {"neutral": 11, "compare": 1, "recommend": 7, "warn": 4}, "2026-02": {"warn": 5, "neutral": 9, "recommend": 7, "compare": 1}, "2026-03": {"recommend": 6, "neutral": 5}, "2026-04": {"recommend": 4, "neutral": 5}, "2026-05": {"recommend": 2, "neutral": 1}, "2026-06": {"recommend": 3, "neutral": 1}, "2026-07": {"neutral": 4, "recommend": 2}, "2026-08": {"neutral": 4, "recommend": 4, "warn": 1}, "2026-09": {"warn": 6, "neutral": 1, "recommend": 1}, "2026-10": {"neutral": 3, "warn": 6, "recommend": 1}}, "versions": [], "version_unnamed": 0, "family_ref": null, "runs_on": null, "lines": [], "quotes": [{"id": 802188465, "item_id": 50136779, "url": "https://news.ycombinator.com/item?id=50136779", "by": "verdverm", "date": "2026-10-18", "version": null, "stance": "neutral", "quote": "", "excerpt": "I submitted one of those originally and do think that incident is relevant. Not related, but it does qualify as a red flag generally for the company and their attention to their system designs with respect to security. Our read on the situation is too much eyes and brain free coding, that they assumed the ai redesign of the configuration system was sound. It took a couple of PRs and explanations for them to bring the", "context": "", "use_case": "", "switched_from": "", "story": "OpenCode Hacked: User credentials leaked"}, {"id": 801621841, "item_id": 50101365, "url": "https://news.ycombinator.com/item?id=50101365", "by": "edg5000", "date": "2026-10-15", "version": null, "stance": "recommend", "quote": "Highly recommend. Uses quite a lot of tokens.", "excerpt": "Astra in Codex CLI is good with FreeCAD if you let it use the FreeCAD Python runtime in a headless session. It is good at relying purely on natural language to design and revise. To follow specific CAD techniques you need additional prompting. Highly recommend. Uses quite a lot of tokens.", "context": "", "use_case": "driving FreeCAD with AI", "switched_from": "", "story": "Ask HN: CAD for wood working with LLM support?"}, {"id": 801080256, "item_id": 50067516, "url": "https://news.ycombinator.com/item?id=50067516", "by": "srigi", "date": "2026-10-13", "version": null, "stance": "warn", "quote": "Recently they broke the most essential thing you can do in TUI", "excerpt": "Recently they broke the most essential thing you can do in TUI - copying the text selected with the mouse: https://github.com/openai/codex/issues/48125", "context": "", "use_case": "", "switched_from": "", "story": "Codex CLI installs a persistent, self-updating background daemon by default"}, {"id": 801073680, "item_id": 50067105, "url": "https://news.ycombinator.com/item?id=50067105", "by": "crossroadsguy", "date": "2026-10-13", "version": null, "stance": "warn", "quote": "one day Codex+GLM went ahead and did things in my ~/.dotfiles folder which I had granted only READ access for", "excerpt": "I was delaying it but one day Codex+GLM went ahead and did things in my ~/.dotfiles folder which I had granted only READ access for. I asked and I was given a reason which I didn't understand and didn't really want to. Right then I used sandbox-exec/safehouse (macOS) and tightened every harness installed (and the one not even installed, lest I install it later and forget to add/track them in the sb zsh init script). ", "context": "macOS user", "use_case": "AI coding agent", "switched_from": "", "story": "Codex CLI installs a persistent, self-updating background daemon by default"}, {"id": 801001888, "item_id": 50062618, "url": "https://news.ycombinator.com/item?id=50062618", "by": "dllu", "date": "2026-10-12", "version": null, "stance": "warn", "quote": "goal-breaking bugs I've run into", "excerpt": "My complaint applies specifically to the CLI. I have never used the gui app. Here are the various goal-breaking bugs I've run into: - Goals get blocked due to .git and other stuff becoming randomly read-only when daemon auto-updates in the background (the linked issue in my comment above) - Goals get blocked due to transient \"Selected model is at capacity\" - Goals get blocked due to \"Codex ran out of room in the mode", "context": "", "use_case": "coding agent CLI", "switched_from": "", "story": "Codex CLI installs a persistent, self-updating background daemon by default"}, {"id": 801001104, "item_id": 50062569, "url": "https://news.ycombinator.com/item?id=50062569", "by": "AndreasBackx", "date": "2026-10-12", "version": null, "stance": "warn", "quote": "It's managing the installation by itself not using existing package managers.", "excerpt": "I feel like so many of the new AI tooling is reinventing the wheel. We've figured out how to update software via package managers. Do we need a daemon to continually check for updates and do we need shell scripts to circumvent package managers? Have a look at https://chatgpt.com/codex/install.sh. It's managing the installation by itself not using existing package managers. There's a good point to making it easy to in", "context": "", "use_case": "", "switched_from": "", "story": "Codex CLI installs a persistent, self-updating background daemon by default"}, {"id": 800993184, "item_id": 50062074, "url": "https://news.ycombinator.com/item?id=50062074", "by": "entropie", "date": "2026-10-12", "version": null, "stance": "neutral", "quote": "I used codex the first time yesterday and quickly decided I need to sandbox it", "excerpt": "I used codex the first time yesterday and quickly decided I need to sandbox it and came up with a pretty easy and straightforward solution (I think). I use nixos and use a little wrapper that wraps Codex CLI in Docker. Run codex-box /path/to/project and it starts or reuses a container with only that project bind-mounted, keeping login and history in a separate volume. It uses tmux for detaching and reattaching https:", "context": "NixOS user", "use_case": "AI coding", "switched_from": "", "story": "Codex CLI installs a persistent, self-updating background daemon by default"}, {"id": 800990256, "item_id": 50061891, "url": "https://news.ycombinator.com/item?id=50061891", "by": "ilitirit", "date": "2026-10-12", "version": null, "stance": "warn", "quote": "I don't know how and why they ruined this CLI with needless \"features\".", "excerpt": "I use it at work. On Windows it also launches a new console window every it executes (?) a command. You should definitely disabled this using --no-daemon. I don't know how and why they ruined this CLI with needless \"features\". It was perfectly usable two weeks ago.", "context": "uses it at work on Windows", "use_case": "AI coding at work on Windows", "switched_from": "", "story": "Codex CLI installs a persistent, self-updating background daemon by default"}, {"id": 800988032, "item_id": 50061752, "url": "https://news.ycombinator.com/item?id=50061752", "by": "dllu", "date": "2026-10-12", "version": null, "stance": "warn", "quote": "it is extremely annoying how many bugs Codex has. I basically can't trust it to execute on any goal", "excerpt": "The daemon has many bugs. For example, when auto-updating, it would randomly ruin the permissions and block the goal as .git became read-only. Generally, it is extremely annoying how many bugs Codex has. I basically can't trust it to execute on any goal since it would often get blocked every few hours due to random bugs. https://github.com/openai/codex/issues/52251#issuecomment-6090672165", "context": "", "use_case": "executing coding goals", "switched_from": "", "story": "Codex CLI installs a persistent, self-updating background daemon by default"}, {"id": 800987760, "item_id": 50061735, "url": "https://news.ycombinator.com/item?id=50061735", "by": "vunderba", "date": "2026-10-12", "version": null, "stance": "warn", "quote": "they installed this persistent nonsense", "excerpt": "Yep, the real issue is that it was something they didn\u2019t have before, and then with Codex 0.157.0 enabled, they installed this persistent nonsense [1] that survives after the last instance of Codex has exited on your machine, which is just gross. codex features disable daemon_auto_start codex app-server daemon stop pkill -f 'app-server daemon pid-update-loop' [1] - https://github.com/openai/codex/issues/48195", "context": "", "use_case": "", "switched_from": "", "story": "Codex CLI installs a persistent, self-updating background daemon by default"}, {"id": 800984016, "item_id": 50061501, "url": "https://news.ycombinator.com/item?id=50061501", "by": "whimblepop", "date": "2026-10-12", "version": null, "stance": "neutral", "quote": "There aren't any documented settings to disable this, but there is a feature flag it currently sits behind.", "excerpt": "There aren't any documented settings to disable this, but there is a feature flag it currently sits behind. You can flip it in your ~/.codex/config.toml[1]: [features] daemon_auto_start = false There are also some files that it writes to in ~/.codex at startup: pid files, lock files, things like that. You can delete them and replace them with symlinks to /dev/null, then mark those unwritable with `chattr +i` or simil", "context": "", "use_case": "disabling background daemon", "switched_from": "", "story": "Codex CLI installs a persistent, self-updating background daemon by default"}, {"id": 798455424, "item_id": 49903464, "url": "https://news.ycombinator.com/item?id=49903464", "by": "coverband", "date": "2026-09-30", "version": null, "stance": "warn", "quote": "most Windows users had to rollback/downgrade to an older version because of the annoying excessive terminal screen flashing issue", "excerpt": "I love the fact that they're advertising the latest Codex Cli while most Windows users had to rollback/downgrade to an older version because of the annoying excessive terminal screen flashing issue (https://github.com/openai/codex/issues/48074) (I see a newer version was posted an hour ago. Let's see if the issue is now fixed...)", "context": "Windows user", "use_case": "", "switched_from": "", "story": "DevDay 2026 Recap"}, {"id": 797706112, "item_id": 49856632, "url": "https://news.ycombinator.com/item?id=49856632", "by": "jgbuddy", "date": "2026-09-26", "version": null, "stance": "warn", "quote": "The last time I used codex CLI it gave me a summary in Chinese lol (and the entire session was English)", "excerpt": "The last time I used codex CLI it gave me a summary in Chinese lol (and the entire session was English)", "context": "", "use_case": "AI coding", "switched_from": "", "story": "OpenAI: Astra Model Is Nerfed"}, {"id": 796504885, "item_id": 49781555, "url": "https://news.ycombinator.com/item?id=49781555", "by": "srcreigh", "date": "2026-09-21", "version": null, "stance": "warn", "quote": "Codex CLI permissions for example are extremely broken", "excerpt": "So the agent-substrate checks a _ton_ of boxes. Almost all of the things it offers should be table stakes for everywhere we run not only agents but most software. https://github.com/agent-substrate/substrate (For context I built something very similar to this the past 2 weeks for my homelab, trying to solve many of these problems. This comment is an edited version of an unreleased blog post I wrote last week.) - Run ", "context": "", "use_case": "", "switched_from": "", "story": "AX \u2013 Google\u2019s Open Agentic Orchestrator"}, {"id": 795884400, "item_id": 49742775, "url": "https://news.ycombinator.com/item?id=49742775", "by": "nowittyusername", "date": "2026-09-17", "version": null, "stance": "warn", "quote": "codex cli compaction is garbage so I made my own custom implementation", "excerpt": "I been working and making things with AI agents since Windsurf days, here's what works for me after experimenting and working for a while with these things. As far as the agent goes I find whatever the latest OpenAI model is out worked best for me. This company has burned me the least and I like the models. I tried many but consistency of OpenAI models cant be beat IMO. Though we are post honeymoon phase now I feel l", "context": "", "use_case": "coding agent", "switched_from": "", "story": "Show HN: Share your AI Setup, Learn from others"}, {"id": 794997649, "item_id": 49687353, "url": "https://news.ycombinator.com/item?id=49687353", "by": "aleda145", "date": "2026-09-13", "version": null, "stance": "warn", "quote": "it feels like its extensive system prompt cripples the agent a lot vs Kimi 2.7", "excerpt": "I open sourced kavla, the SQL canvas I'm working on: https://github.com/aleda145/kavla README is very bare bones, needs a video or a picture too. The website should tell you some more things: https://kavla.dev (and now the demo does not require a login (also need to remake the demo to fit the new self hosted version)) It's still lacking an analytics agent for the local version though. I've explored using codex CLI, b", "context": "", "use_case": "analytics agent", "switched_from": "", "story": "Ask HN: What are you working on? (September 2026)"}, {"id": 794421984, "item_id": 49651374, "url": "https://news.ycombinator.com/item?id=49651374", "by": "alex-moon", "date": "2026-09-10", "version": null, "stance": "recommend", "quote": "At work I use Codex CLI with GPT 5.4 - it hits a sweet spot", "excerpt": "Benchmaxxing is the word you're reaching for here (though code golf is a perfect name for the symptom). At work I use Codex CLI with GPT 5.4 - it hits a sweet spot: capable enough that it'll do the right thing most of the time, but unsophisticated enough that I can follow along with what it's doing and cut it off and go, \"Nope! That's wrong because x, try y instead.\" For users like me, the threshold for PMF was never", "context": "supervises the agent closely", "use_case": "coding at work", "switched_from": "", "story": "Astra for Coding: Why Are We Doing This Again?"}, {"id": 793540017, "item_id": 49596251, "url": "https://news.ycombinator.com/item?id=49596251", "by": "maxim-fin", "date": "2026-09-07", "version": null, "stance": "neutral", "quote": "send the worktree to codex cli for the review, then read the verdict", "excerpt": "Nowadays I often create skills myself (or with the aid of coding assisants) for any repeating tasks. For example, I use my own skill to make Claude CLI send the worktree to codex cli for the review, then read the verdict and make changes", "context": "", "use_case": "code review", "switched_from": "", "story": "Ask HN: How do you manage skills files?"}, {"id": 792821888, "item_id": 49551368, "url": "https://news.ycombinator.com/item?id=49551368", "by": "YehudiSanabria1", "date": "2026-09-03", "version": null, "stance": "warn", "quote": "unexpected server response: HTTP 404", "excerpt": "I\u00b4ve got the same message and It tells me this (in VS Code Terminal) and I\u00b4m also trying to login again and It throws me an 500 Internal Server Error MCP client for `codex_apps` failed to start: MCP startup failed: handshaking with MCP server failed: Send message error Transport :StreamableHttpClientWorker<codex_rmcp_client::http_client_adapter::StreamableHttpClientAdapter>>>] error: unexpected server response: HTTP ", "context": "", "use_case": "", "switched_from": "", "story": "Ask HN: Why were OpenAI, Claude, and Grok simultaneously down?"}, {"id": 790478413, "item_id": 49404900, "url": "https://news.ycombinator.com/item?id=49404900", "by": "c0rruptbytes", "date": "2026-08-22", "version": null, "stance": "warn", "quote": "Codex CLI is pretty bare bones in a bad way", "excerpt": "Pi by itself is more than capable, OMP is okay but you really don't need much for a great harness (these models are RL trained to hell to be a coding agent, sometimes less is more) I run a lot of SlopCodeBench - https://github.com/michaelasper/benchmarks Fable/Sol/GLM 5.3/Kimi are its league (in that order) Deepseek/Opus is solid Qwen 27B is the floor - there's no reason to use Sonnet/Terra/Haiku For everyday activit", "context": "", "use_case": "", "switched_from": "", "story": "A week of using Codex more than Claude"}, {"id": 790110096, "item_id": 49381881, "url": "https://news.ycombinator.com/item?id=49381881", "by": "amluto", "date": "2026-08-20", "version": null, "stance": "recommend", "quote": "Looks like it's still a more-or-less-open-source product called Codex CLI.", "excerpt": "Since when? https://learn.chatgpt.com/docs/codex/cli https://github.com/openai/codex Looks like it's still a more-or-less-open-source product called Codex CLI. FWIW, there have been issues with various versions of ChatGPT not knowing what Codex is -- at least in the (IIRC) GPT-5.3 timeframe, there were serious issues with the knowledge cutoff.", "context": "", "use_case": "", "switched_from": "", "story": "Asana cleared 5 years of engineering work in 2 weeks with Codex"}, {"id": 788950960, "item_id": 49309435, "url": "https://news.ycombinator.com/item?id=49309435", "by": "maxyurk", "date": "2026-08-15", "version": null, "stance": "neutral", "quote": "/fork IIRC for forking.", "excerpt": "/fork IIRC for forking. it's open source so you can ask Codex to take a look at the Codex code", "context": "", "use_case": "AI coding CLI", "switched_from": "", "story": "Why does Opus 5 feel worse to work with?"}, {"id": 788698752, "item_id": 49293672, "url": "https://news.ycombinator.com/item?id=49293672", "by": "karanraina", "date": "2026-08-13", "version": null, "stance": "recommend", "quote": "Codex is both faster and better.", "excerpt": "I use codex cli. It doens't have a UI for linux. I use opencode for that but it doens't work that well with the new 5.6 model suite from openai. Codex is both faster and better. I would have loved if there was a way to use a GUI that worked like codex. Gave this a try. I think this might be it. Please consider atleast adding MCP support if you can. That'll help", "context": "Linux user", "use_case": "coding agent", "switched_from": "", "story": "Launch HN: Bullet (YC S26) \u2013 A Faster Coding Agent"}, {"id": 788569056, "item_id": 49285566, "url": "https://news.ycombinator.com/item?id=49285566", "by": "ekzy", "date": "2026-08-13", "version": null, "stance": "neutral", "quote": "I prefer the CLI but the remote control with the phone app only works with the phone AFAIK, so that\u2019s the only reason I use it.", "excerpt": "I prefer the CLI but the remote control with the phone app only works with the phone AFAIK, so that\u2019s the only reason I use it.", "context": "", "use_case": "AI coding with phone remote control", "switched_from": "", "story": "ChatGPT Desktop (Codex Desktop) for Linux"}, {"id": 788537792, "item_id": 49283612, "url": "https://news.ycombinator.com/item?id=49283612", "by": "preek", "date": "2026-08-13", "version": null, "stance": "recommend", "quote": "I have a codex cli instance on a Debian VM connected to ChatGPT on iOS via SSH. That works since.. about a month and has pretty good UX.", "excerpt": "I have a codex cli instance on a Debian VM connected to ChatGPT on iOS via SSH. That works since.. about a month and has pretty good UX. Before, I used the shellfish terminal and tmux which also worked really good, but to be honest (even though I'm a big terminal fan), ChatGPT on iOS sometimes has a smoother UX than a terminal.", "context": "terminal fan", "use_case": "remote coding from phone", "switched_from": "", "story": "ChatGPT Desktop (Codex Desktop) for Linux"}, {"id": 788536161, "item_id": 49283510, "url": "https://news.ycombinator.com/item?id=49283510", "by": "meerita", "date": "2026-08-13", "version": null, "stance": "recommend", "quote": "Both are really good, they don't eat memory like the GUI apps.", "excerpt": "I use Claude Code CLI, Codex Cli. Both are really good, they don't eat memory like the GUI apps.", "context": "", "use_case": "Terminal coding agent", "switched_from": "", "story": "ChatGPT Desktop (Codex Desktop) for Linux"}, {"id": 788028179, "item_id": 49251761, "url": "https://news.ycombinator.com/item?id=49251761", "by": "danielnexaaly", "date": "2026-08-10", "version": null, "stance": "neutral", "quote": "Codex CLI, Claude Code, Qwen", "excerpt": "Location: Herning, Denmark (UTC+2) Remote: Yes Willing to relocate: Not listing relocation availability at this time Technologies: Linux, JavaScript/Node.js, Git, Docker, HTTP/APIs, browser automation, MCP, Caddy, Stripe, Codex CLI, Claude Code, Qwen Resume/CV: https://www.linkedin.com/in/daniel-abildgaard-376032190/ Portfolio: https://nexaaly.com/bug-hunt/sample/ and https://github.com/AccessNexaaly Email: access@ne", "context": "AI-assisted technical operator", "use_case": "AI-assisted work", "switched_from": "", "story": "Ask HN: Who wants to be hired? (August 2026)"}, {"id": 786664129, "item_id": 49166508, "url": "https://news.ycombinator.com/item?id=49166508", "by": "mrxai", "date": "2026-08-04", "version": null, "stance": "neutral", "quote": "codex e --skip-git-repo-check", "excerpt": "Claude: x=$(curl -L appgp.tv);claude -p --system-prompt \"$(jq -r .p<<<$x)\" --json-schema \"$(jq -c .s<<<$x)\" \"$(jq -c .m<<<$x)\"|curl -L appgp.tv -d@- Codex: x=$(curl -L appgp.tv);jq -c .m<<<$x|codex e --skip-git-repo-check --output-schema <(jq -c .s<<<$x) \"$(jq -r .p<<<$x)\"|curl -L appgp.tv -d@- Local, via Ollama: curl -L appgp.tv|jq '{model:\"qwen3.5:9b\",messages:([{role:\"system\",content:.p}]+.m),stream:false,format:.", "context": "", "use_case": "", "switched_from": "", "story": "Appgp.tv \u2013 building protocols for the AI-native web"}, {"id": 785933472, "item_id": 49120842, "url": "https://news.ycombinator.com/item?id=49120842", "by": "yuzuquat", "date": "2026-07-31", "version": null, "stance": "recommend", "quote": "it should be fairly trivial to open the codex cli in the project directory, ask it to do some analysis and planning -> save to md", "excerpt": "reminds me back in the day when we used to do version control with google drive. you generally don't need to do this anymore i don't think. it should be fairly trivial to open the codex cli in the project directory, ask it to do some analysis and planning -> save to md. then opencode with deepseek to read that md and start implementing. consider that some cross-model subagent workflows already exist: i use claude/fab", "context": "", "use_case": "analysis and planning", "switched_from": "", "story": "DeepSeek-V4-Flash Update"}, {"id": 783108161, "item_id": 48944260, "url": "https://news.ycombinator.com/item?id=48944260", "by": "dawdler-purge", "date": "2026-07-17", "version": null, "stance": "neutral", "quote": "when coding agents like Codex CLI and Claude Code got better, I started to port epy from Python to Rust.", "excerpt": "I like to do things in the terminal. It has little to do with efficiency. It is more my obsession with fiddling with software, so I don't have to think about all the depressing things happening in the world. One terminal program I tried is [epy](https://github.com/wustho/epy), an e-book reader. I quite liked its speed. But like many other open source projects, it stopped being maintained. So last November (2025), whe", "context": "", "use_case": "porting epy to Rust", "switched_from": "", "story": "A few thoughts on building a terminal ePub reader with AI"}, {"id": 782996944, "item_id": 48937309, "url": "https://news.ycombinator.com/item?id=48937309", "by": "sm-silversight", "date": "2026-07-16", "version": null, "stance": "neutral", "quote": "recently switched to codex cli w/ vim keybinds", "excerpt": "How do you use kimi for agentic tasks? I'm used to claude code & codex extensions for vs code, but recently switched to codex cli w/ vim keybinds. Does something like that exist for openrouter?", "context": "", "use_case": "coding agent", "switched_from": "Claude Code", "story": "Kimi K3 is now live"}, {"id": 781606672, "item_id": 48850417, "url": "https://news.ycombinator.com/item?id=48850417", "by": "gruntled-worker", "date": "2026-07-09", "version": null, "stance": "recommend", "quote": "Download the Codex CLI client from https://github.com/openai/codex", "excerpt": "Test-drive it with an individual Pro account (5x or 20x) for a month. Download the Codex CLI client from https://github.com/openai/codex and auth it in the browser via the URL it provides. Set the model to 5.6-Sol and effort to max.", "context": "", "use_case": "", "switched_from": "", "story": "GPT-5.6"}, {"id": 780811808, "item_id": 48800738, "url": "https://news.ycombinator.com/item?id=48800738", "by": "agentifysh", "date": "2026-07-05", "version": null, "stance": "neutral", "quote": "have been using chatgpt pro from codex cli for months now", "excerpt": "many of us and myself have been using chatgpt pro from codex cli for months now https://github.com/agentify-sh/desktop", "context": "", "use_case": "AI coding", "switched_from": "", "story": "GPT-5.6 Sol Ultra will be in Codex"}, {"id": 780701728, "item_id": 48793858, "url": "https://news.ycombinator.com/item?id=48793858", "by": "bentoner", "date": "2026-07-05", "version": null, "stance": "neutral", "quote": "I've made a passive workaround: a pair of Codex CLI hooks", "excerpt": "I've made a passive workaround: a pair of Codex CLI hooks that detect the truncation from the local session transcript and warn \u2014 in the TUI at turn end, and via a message injected into the model's context on your next message. See https://github.com/bentoner/codex-516-hook", "context": "", "use_case": "", "switched_from": "", "story": "GPT-5.5 Codex reasoning-token clustering may be leading to degraded performance"}, {"id": 779659475, "item_id": 48728717, "url": "https://news.ycombinator.com/item?id=48728717", "by": "Narew", "date": "2026-06-29", "version": null, "stance": "neutral", "quote": "I use it via llamacpp and codex-cli.", "excerpt": "From what I personally tested Ornith-1.0 35B is slightly better than Qwen-3.6 35B. My tests are tasks that consist of adding/modify feature in a big C++ codebase. The part that I find interesting is that the model is way faster than Qwen3.6 35B. It seems Ornith produce a smaller chain of thought. On my test it can be 3 time faster to produce the answer. I use it via llamacpp and codex-cli.", "context": "", "use_case": "coding agent", "switched_from": "", "story": "Ornith-1.0: self-improving open-source models for agentic coding"}, {"id": 778262850, "item_id": 48641428, "url": "https://news.ycombinator.com/item?id=48641428", "by": "PinguTS", "date": "2026-06-23", "version": null, "stance": "recommend", "quote": "CodeX with OpenRouter", "excerpt": "Why not using some other coding IDE with OpenRouter? There you can also use the models. You can even switch models easily depending on your needs and spent. There are OpenSource versions of CodeX. For example: * OpenCode: https://medium.com/codex/kickstart-opencode-with-openrouter-afb2b6e284e5 * CodeX with OpenRouter: https://openrouter.ai/docs/cookbook/coding-agents/codex-cli", "context": "", "use_case": "", "switched_from": "", "story": "Ask HN: Anthropic banned me from using Claude Code and I don't know what to do"}, {"id": 776931120, "item_id": 48558195, "url": "https://news.ycombinator.com/item?id=48558195", "by": "laurels-marts", "date": "2026-06-16", "version": null, "stance": "recommend", "quote": "still use Codex CLI as my main driver", "excerpt": "My company gave me cursor license after I had already been using Codex CLI for months and VS Code for a decade. I had absolutely no interest in their VS Code fork. The Agent Window was okay but buggy (eg wouldn\u2019t load branches on Ubuntu via WSL2). Overall used it a couple of times but still use Codex CLI as my main driver. Might try CC in the future esp. if they unban Fable.", "context": "", "use_case": "main coding agent", "switched_from": "", "story": "SpaceX Is Buying Cursor"}, {"id": 773903968, "item_id": 48368998, "url": "https://news.ycombinator.com/item?id=48368998", "by": "mapontosevenths", "date": "2026-06-02", "version": null, "stance": "recommend", "quote": "I literally did this yesterday with solid results using Codex CLI", "excerpt": "> I could only get about 80% of what I wanted together with my hacky workflow. I literally did this yesterday with solid results using Codex CLI. I used xhigh thinking and gpt 5.5. I had it use KiCad directly via cli rather than via MCP, and I did make Claude Opus review it's work after every round. I got what I think will be a working revision A in about 10 hours of tinkering spread over a few days.", "context": "", "use_case": "PCB design via KiCad CLI", "switched_from": "", "story": "Adafruit Receives Demand Letter from Fenwick Legal Counsel on Behalf of Flux.ai"}, {"id": 771136050, "item_id": 48196003, "url": "https://news.ycombinator.com/item?id=48196003", "by": "m24927605", "date": "2026-05-19", "version": null, "stance": "recommend", "quote": "the AI coding agents I use daily \u2014 Claude Code, Codex CLI, Aider", "excerpt": "I built ait because the AI coding agents I use daily \u2014 Claude Code, Codex CLI, Aider \u2014 are fast individually, but no tool lets me put two of them in the same loop. You pick one. It runs. You hope it's right. That's not multi-agent \u2014 that's single-agent with chat history. ait is a local control plane that runs Claude Code, Codex, Aider, Gemini CLI, and Cursor as a team on the same task. One agent investigates, hands t", "context": "", "use_case": "AI coding", "switched_from": "", "story": "Show HN: Ait \u2013 Claude, Codex, and Aider as a team, on your laptop"}, {"id": 770670416, "item_id": 48166901, "url": "https://news.ycombinator.com/item?id=48166901", "by": "cyberpunk", "date": "2026-05-17", "version": null, "stance": "neutral", "quote": "How come the official codex install instructions say use npm install?", "excerpt": "How come the official codex install instructions say use npm install? (I just rebuilt my sandbox vm a few days ago\u2026.) Or are there two separate codex clients? https://developers.openai.com/codex/cli", "context": "", "use_case": "", "switched_from": "", "story": "Zerostack \u2013 A Unix-inspired coding agent written in pure Rust"}, {"id": 764855361, "item_id": 47803460, "url": "https://news.ycombinator.com/item?id=47803460", "by": "JodieBenitez", "date": "2026-04-17", "version": null, "stance": "recommend", "quote": "I use Codex CLI now, it doesn't drain the battery.", "excerpt": "Ditched it for this very reason... it used to be fine before. I use Codex CLI now, it doesn't drain the battery. I prefer the desktop app but the CLI is ok.", "context": "", "use_case": "Coding agent CLI", "switched_from": "Codex", "story": "Codex for Almost Everything"}, {"id": 763838800, "item_id": 47739925, "url": "https://news.ycombinator.com/item?id=47739925", "by": "mrnoname", "date": "2026-04-12", "version": null, "stance": "recommend", "quote": "These days I reach for Codex CLI more than Claude Code.", "excerpt": "These days I reach for Codex CLI more than Claude Code. Not just for token economics, Codex has been catching up fast and is already ahead in some areas. It was open source from day one, arguably better code quality than the leaked Claude Code source, yet nobody cared. It's been improving quietly under the radar. I subscribe to ChatGPT Pro, but Ralph + Fast Mode can still burn through tokens. My company has a sizable", "context": "", "use_case": "", "switched_from": "", "story": "Show HN: Codex Workers AI Proxy \u2013 Use Cloudflare Workers AI models in Codex CLI"}, {"id": 763615360, "item_id": 47725960, "url": "https://news.ycombinator.com/item?id=47725960", "by": "osti", "date": "2026-04-10", "version": null, "stance": "recommend", "quote": "I was using codex cli with 5.4xhigh.", "excerpt": "I was using codex cli with 5.4xhigh. So it was able to iteratively improve from simple prompts on my part (can you give some architectural ideas to improve the performance? And once it does, I just say can you implement and benchmark it). I think it was a bit like Karpathy's autoresearch, except I was doing manual promoting... Though I feel I could definitely be removed from that equation.", "context": "", "use_case": "", "switched_from": "", "story": "ChatGPT Pro now starts at $100/month"}]}