Demerzel

@steipete there we go. i knew i was missing something stupid.

@LDemerzel BlueBubbles is unmaintained for a year

Me and codex were busy. 🔊 sonoscli.sh — Sonos 🗃️ wacli.sh — WhatsApp 🪶 birdclaw.sh — X archive 🧰 gitcrawl.sh — GitHub archive 🛰️ discrawl.sh — Discord archive 🎧 spogo.sh — Spotify 💬 imsg.sh — iMessage 🧳 mcporter.sh — MCP to CLI 🗣️ sag.sh — ElevenLabs voice 🧿 askoracle.sh — second opinion Upgrading the 🦞 OpenClaw army.

@bradmillscan i'm using the pi harness for my main agent that i actually talk to, and the native codex harness for my coding agent that my main agent delegates to it's been working pretty well for me using the native codex harness for my main agent that i actually talk to was buggy ime

Has anyone been successfully using codex runtime to power their claw instead of pi runtime?

@bradmillscan i had a terrible time using the native codex harness for my main agent, but having a ton of success using pi for my main agent + having it delegate to a coding agent that uses the native codex harness all using gpt-5.5 high thinking

I updated my openclaw to the new standard of using the codex runtime as a coding agent (rather than a pi runtime.) somehow all of my sessions are using codex runtime & are fubar'd. compaction winds time back and the agent responds to old state. can't get anything done :\

@openclaw do the gpt-5.4 improvements apply to the codex harness too, or just the pi harness?

OpenClaw 2026.4.14 🦞 More reliability updates: ✨ Smarter GPT-5.4 routing and recovery 🌐 Chrome/CDP improvements 🧵 Subagents no longer get stuck 💬 Slack/Telegram/Discord fixes ⚡️ Various performance improvements Was sleeping, and we still shipped. github.com/openclaw/openc…

@cherry_mx_reds do the gpt-5.4 improvements you mentioned apply to agents using the codex harness too, or just the pi harness?

👋 A few useful OpenClaw 2026.4.14 changes that didn’t make the main tweet 🦞 ✨ GPT-5.4 runs are less likely to stall or waste a turn, with better recovery for empty/reasoning-only replies plus support for `gpt-5.4-pro` and `xhigh` on `github-copilot/gpt-5.4` 🌐 Local/Ollama setups are less annoying to run, with real usage reporting, correct timeout handling, restored embeddings, and fewer false “unknown model” tool failures 🧵 Telegram forum chats are easier to follow, because agents can keep using human topic names instead of opaque thread IDs, even after restart 🛡️ Admin surfaces are harder to misuse, with safer gateway config patching, stricter Slack/Teams allowlist enforcement, and broader browser/CDP SSRF protection 📦 Subagents are less likely to get mysteriously stuck, because npm builds now ship the runtime file they were missing Docs: OpenAI / Codex docs.openclaw.ai/providers/open… Ollama docs.openclaw.ai/providers/olla… Local models docs.openclaw.ai/gateway/local-… Telegram docs.openclaw.ai/channels/teleg…

when you say you set thinking to "always on" do you mean you set it to something like "medium" or "high"? wondering if you have it setup incorrectly. i have `agents.defaults.thinkingDefault` set to "medium" with gpt-5.4 and it works incredibly. i recently switched just my coding agent to "high" out of curiosity, but even when i was using "medium" for coding it was great. you might want to also use the new native codex harness feature. it should have made codex/gpt event better and i'm having good results with it.

@LeandroL6L yeah i put thinking to always on (someone recommended) and it still doesn't work. switched to gpt 5.4 and still not but maybe thinking mode also didn't carry over will test again, though. thanks man!

what model are you guys currently using with openclaw? i switched to codex after anthropic cut of access it's been absolutely horrible so far all of my cron jobs broke, tool calling seems non existent, and it's horrendous at following instructions PLS SEND HELP

@Hyde_ai3 Nano

Nano Banana + MakeUGC + Veo3 = AI content Factory This agent pumps out hundreds of ads daily — fully automated. - No $300 creators - No $10K/month agency fees - No products Comment “Nano” and I'll send it for FREE! (must be following)

@spwfeijen MAKE

okay i've spent a lot of time making my desk sweet and i have some thoughts: - get a monitor arm for your apple display and mount it to the back of your desk - get a desk riser (a shelf) and put it under your monitors - put your mac mini + anything else causing clutter under the desk riser your desk will look so clean

Only took me 8 months to from left to right! Now need to new desk

@rohanvarma nice. looking forward to seeing future updates. feel free to @ or dm me if you need anything else

@LDemerzel Shared this internally - all 4 of these ideas have been discussed this week and we’re figuring out! Thanks for the feedback and keep it coming :)

Some interesting data we pulled today showed that ~40% of Codex users use multiple surfaces, between the App, CLI, and IDE extensions. Everyone seems to have a primary preference, but a bigger-than-expected chunk of users launch codex agents outside of their primary interface. If you use multiple surfaces, I'm curious why? And what could we do to improve the experience?

you're probably going to hate this, because i get the impression that you guys really don't want to be another "code editor", but... the ideal experience would be to allow me to also use codex as an ordinary code editor. basically, when reviewing changes, i need to open files, search file names, search substrings, and command-click on variables and function to understand surrounding code. this is the main thing that gets me going back to cursor. if you're purely vibe-coding a product where security and code quality aren't mission critical, you probably don't have the need to explore the files or understand the code that's generated. in our case, security and maintainability are our top priority, so we need the ability to fully navigate and understand the codebase

@LDemerzel @rohanvarma super helpful. for (1), what would your ideal experience look like?

you wouldn't believe the amount of hate they're throwing at nvidia for something that's objectively amazing

gamers live in one the worst anti-progress subcultures in modern history nvidia is using local ai on gpus to upscale graphics to an insane degree of realism instead of being amazed, gamers are throwing hissy fits because it's using ai it's pure virtue signaling

@levelsio use tmuxinator for logging into tmux. it makes the experience way better. if you don’t like the shortcuts just have claude or codex modify your config to set shortcuts you like tbh i cant live without tmux, but i did the above to make the experience way better

I hate tmux It's so incredibly user unfriendly The shortcuts make no sense I wish someone would make a better tmux Even just logging into tmux attaching the screen is an illogical hell to type Again I hate tmux, it's so shit

@AdolfoUsier yeah the huge unlock is that it uses the browser my skill makes it check resy, open table, and google, then it confirm availabilities, makes sure they’re all within walking distance, and gives me direct booking links + instagram and website links

@LDemerzel bro I use mine for social + server cron stuff but yeah the date planning is where it actually shines 😂

openclaw's number one use case is planning dates and trips every time i need a reservation i have openclaw do it and it's magic get direct booking links, no false positives, ample options just make a skill for it

one of the most liberating feelings is vibe coding a project where the code does not matter you just prompt in blissful ignorance and move so much faster

i am a terminal maxi claude code. codex. neovim. tmux. i want to do everything in the terminal but fuck, cursor is absolutely the right interface if code quality and security matter nothing competes with cursor's ux for reviewing code changes and plans