Shreyansh Lodha @staticvar_dev

I was promised it will be fixed in a month. But it is still not! Clearly looks like they have intentionally kept this dark pattern! #shame @we_are_cult

I’ll never understand why do I have to unpause my membership today(4th) to book a session on 8th? @we_are_cult this is such a basic user flow. Absolutely reckless and unprofessional. #scam #cultfit

I think @thsottiaux and team are cooking something, my weekly drain rate is right now close to 5.5 levels…

x.com/i/article/2077…

I had my first real “wow” moment with gpt 5.6 sol (high) yesterday. I told Codex: “Fix Honcho’s recall issue. I’m going to bed.” That was it. I didn’t even use /goal. For context, Honcho had stopped processing newer conversations after my local LLM routing key expired. Records were still being marked “processed,” but recall for newer conversations was failing. While I slept, Codex did all of this: - Discovered that 7K+ processed records hadn’t fully become usable memories - Fixed the LLM routing and replayed the historical data - Caught Gemini’s embedding quota exhaustion and correctly refused to mix old incompatible embedding models - Deployed Qwen through Ollama on my ARM64 VPS - Re-embedded old ~16K messages and documents - Resumed the stalled migration actual wow starts here... It autonomously created an hourly scheduled monitor, basically using codex's cronjob / heartbeat to track progress and fix issues along the way. It actually waited for zero pending or failed records and then tested actual end-to-end memory recall. Once everything passed, it disabled the migration worker, and deleted its own monitoring job after 20+ scheduled checks and multiple context compactions. The whole thing ran unattended overnight and into the next day. Doing this myself would have taken 3–4 hours just to diagnose and deploy the fix and then monitor the migration. But... YOU CAN JUST DO THINGS NOW!

Hi everyone 👋 I’m exploring new Senior / Staff / Principal Android engineering opportunities. If you know of a role that could be a fit, I’d really appreciate an intro or a quick chat (DMs are open). A little about me: • 14+ years in native Android across Equal AI, JioCinema (Viacom18), Gojek, Meesho, and DoorDash - products used by 100M+ people • Google Developer Expert for Android since 2021. Built Android from zero to 450K DAU at Equal AI, led the Core Mobile Platform team at JioCinema through IPL-scale traffic, and shipped weekly KMP libraries powering Android, iOS, and Web •Deep in Kotlin, Compose, KMP, and mobile platform engineering - plus agentic AI tooling: MCP servers, spec-driven development, and open-source work like compose-skill (480+ ⭐), Andromeda, and ComposeProoof •Comfortable owning technical direction end to end, from architecture strategy to production, while staying hands-on in the code I’m open to remote contract work (worldwide) and permanent Staff/Principal roles in Europe with relocation support. P.S. If your intro leads to my next role, I’ll gift you lifetime access to the agentic mobile tooling I’m building.

Here’s what a custom TOML template could look like. It means less repetition, and importing a template could handle most of the “how should I prompt this model?” problem. You could have different templates for different tasks like PR reviews, bug fixes, research, planning, and so on. I know a lot of this is already possible with skills or AGENTS.md, but this could offer more control and flexibility through hook integration and the ability to override certain config.toml properties. Overall, it would give prompts more structure and hopefully lead to more reliable, predictable results.

After reading this I felt that the entire process of learning how to prompt a new model should require less thought, especially given the release cadence of models these days. So what if codex had "prompt templates"? @ajambrosino @thsottiaux @jxnlco

eric provencher @pvncher

3 weeks ago

With 5.6 Sol, a lot of people are still prompting the model exactly as they did 5.5 It's important to note that 5.6 Sol is a lot more tenacious and thorough than previously models. Check out the guide I wrote here to get better outcomes learn.chatgpt.com/docs/prompting

67 235 3K 354K 4K

@thsottiaux @flyosity Can we fix this too while we are at it?

whenever we post any feature - x already does this, you copied it - i haven't heard of x - i look into x, it doesn't do it can you guys just enjoy the tools you use and stop being so lame with it

@airkatakana Even LLMs get bored during Gradle builds!

@Dimillian Whenever I am working with devbox from codex remote (connected via mac), I have to keep the chatgpt app open on screen to make it work, the moment I switch apps it gives "devbox is reconnecting".

I build Android that doesn't page you at 3am. 14 yrs, GDE. 0 → 450K DAU on my last app; KMP at 100M+ scale before that. Lately building agent tooling for Compose (ComposeProof.dev) and github.com/aldefy/compose… github.com/aldefy/compose… Taking on a remote contract or two Principal/Staff. Hiring, or know someone who is? Tag them 👇

I don't get it, how is my usage nil and my 5 hour usage is 0%? I've barely used codex today and I am on 5x Pro plan. Something is definitely off, I've been on 5x Pro plan for a couple of months now and nothing like this has ever happened. @reach_vb @thsottiaux #codex #openai

Trying to submit a build-brief plugin to openai/plugins, but external PRs are not being accepted right now. Everything is ready, validation passes! Anyone from OpenAI or familiar with the plugin marketplace intake able to point me to the right path? @ajambrosino #openai #codex

Hey @INDmoneyApp @INDstocksApp Why can’t I see my US stocks? Everything in the app works perfectly fine. But the US stocks are missing for over a week. I thought it was a bug initially. Need some clarification here! Build V6.6.0

I have to push back on 2 things as i think one is categorically incorrect and the other is demonstrably incorrect. 1. Debugging: Debugging is not a thing if coding is solved. You would produce correct behaviors. I don't understand how a solved problem could produce erroneous behavior. 2. Coding is the easy part: setting hardware, capacity, talking to users, product planning agreed is in fact hard, but so is coding. Example: If coding was in fact not hard then Claude Code having a flickering issue for well over 9 months, which is a purely software challenge, would have been solved almost immediately (immediately being on a shortened time scale comparatively to a human solve time scale). For more trivial applications software approximation can largely work. I also love software approximation for exploring how things should feel.

The team is doing incredible work and this is probably one off issue, I understand!

2. In codex app the output of model is invisible/hidden sometimes. All I see is "Worked for (xx)s" but no output. I have to quit & restart codex and then all of a sudden the output is visible. Couldn't capture the screenshot for this but I was on Version 26.527.31326 (3390)

@ajambrosino @thsottiaux Ran into some problems with Codex app & GPT 5.5 (FYI I did not run into this before today.) 1. Agent stops randomly. (can provide feedback ID if its useful for debugging)