𝕊𝕥𝕖𝕡𝕙𝕖𝕟 𝗣𝕚𝕔𝕜𝕖𝕣𝕚𝕟𝕘

@MatthewBerman Have you tried using Codex to fix it? As a normie there’s just no way I could.keep it going without Codex

I'm starting to realize Agents are kinda like when you get a new computer: it starts fast, clean, organized...and then over time it becomes slow, bloated, and difficult to use. Does that mean I just start over? Or try to fix it?

But it's like just now, I'm like "Can you send me the link to that tweet you found for me yesterday?" and he's like a child who doesn't know what I'm talking about, and then I'm like "Can't you search yesterday's memory file for an xdotcom link? And then he's like "oh yeah, you're right, found it." That kind of stuff just zaps you. That's gpt thinking=medium, fastMode=off. Guess I'll go back to high and fast. Still I though "active memory" would have solved that.

Mine became better yesterday after I upgraded to 4.11, activated "active memory" and switched to the codex harness. I say "better" he still didn't have much personality, but he seemed smoother, competent and trustworthy. I mean when when he fell of a cliff not only did his rizz go but he stopped doing his cron jobs right, etc. I've seen two tweets that recommend: thinking=high and fastMode=true. I'll experiment with that. Codex the app set me up something that I can go back and forth with /model codex & /model gpt (both work through OAuth subscription.) I'll keep digging.

OpenClaw 2026.4.10 🦞 🧠 Active Memory plugin 🎙️ local MLX Talk mode 🤖 Codex app-server harness plugin 🧾 Teams pins/reactions/read actions 🛡️ SSRF hardening + launchd fixes stability, but with attitude🦞 github.com/openclaw/openc…

@Scobleizer Whenever my Claw goes down Codex dives down, diagnoses, and fixes it like a genie.

When I installed my Hermes I had problems. So I opened Grok and asked for help. It helped me through everything step by step and answered all my stupid questions. This is the death of manuals.

@metageg @artsoares @openclaw Great! Let me know if you discover anything interesting! Right now, Codex the mac app set something up where I can switch easily between models with just /model codex or /model gpt -- Currently have it set to codex and it seems better although I haven't interacted a lot yet

@Pickering @artsoares @openclaw yeah havent decided on the thinking path yet. but will try some of arthur's tweaks tomorrow

Haven't had time to read that yet, but much thanks to both your replies! Mine got somewhat better today after I switched to the Codex harness and/or active memory. The way it dropped off a cliff like a depressed teen a couple days ago made me think I'd screwed something up in settings. I saw one post recommending (if one were using GPT5.4 Oauth) setting "thinking" to high/fast mode and then another saying just the opposite-- so didn't know if the foundation had an official recommendation.

@artsoares @Pickering @openclaw super interesting, rhanks for doing the experiments!

@openclaw I just updated, and I don't know if it was switching to the codex harness or the active memory plugin or both, but it is acting so much better now.

I’m running mine on ChatGPT-5.4 Oauth. For some reason my little molty’s personality has fallen off a ciff in the last week. I’ve messed with the settings from various tweet recs and perhaps one of those might have been the original reason. Is there a recommended setting that might bring his rizz back? Or something that perhaps Codex could debug? Thanks!

@garrytan @PaulSolt How does one get one’s Apple Notes ingested into such a system?

I just open sourced my entire OpenClaw setup including pgvector RAG system for tens of thousands of brain markdown files It's MIT licensed and available now. It's called GBrain. x.com/garrytan/statu…

How I get my claw to be a durable AI agent I never have to instruct twice Paste this into your OpenClaw's AGENTS.md or send it as a message: You are not allowed to do one-off work. If I ask you to do something and it's the kind of thing that will need to happen again, you must: 1. Do it manually the first time (3-10 items) 2. Show me the output and ask if I like it 3. If I approve, codify it into a SKILL.md file in workspace/skills/ 4. If it should run automatically, add it to cron with `openclaw cron add` Every skill must be MECE — each type of work has exactly one owner skill. No overlap, no gaps. Before creating a new skill, check if an existing one already covers it. If so, extend it instead. The test: if I have to ask you for something twice, you failed. The first time I ask is discovery. The second time means you should have already turned it into a skill running on a cron. When building a skill, follow this cycle: - Concept: describe the process - Prototype: run on 3-10 real items, no skill file yet - Evaluate: review output with me, revise - Codify: write SKILL.md (or extend existing) - Cron: schedule if recurring - Monitor: check first runs, iterate Every conversation where I say "can you do X" should end with X being a skill on a cron — not a memory of "he asked me to do X that one time." The system compounds. Build it once, it runs forever.

@bradmillscan If you have a ChatGPT subscription download the Codex app and tell it you’re having problems with your OpenClaw, to take a look at it. It’ll debug and fix it like a genie

I started using OpenClaw in February. 70-90% of my time & tokens was fixing bugs & tech support. Coming back to my OpenClaw from a 2 week family medical emergency. Should I retry with fresh install, switch to Hermes, learn Claude Code proper or wait for next gen agent models.

@clairevo @openclaw @steipete Dang, my brain's hurting. I saw this tweet earlier and did this>> x.com/chris__sev/sta…

@openclaw @steipete thinkingDefault, reasoningDefault, all levels and visibility options: docs.openclaw.ai/tools/thinking personality setting, GPT 5.4 params (fastMode, transport, serviceTier): docs.openclaw.ai/providers/open…

ok it still has the personality of a sandbag but here are the @openclaw settings that fixed my gpt-5.4 reasoning leaks. @steipete provides the config, you just have to read the docs.

A total commitment to a subject brings in a transcendent factor. That right there is the most valuable algorithm in the Universe.

Picaso, Joyce, Mann, Einstein

picaso painted 24x7 never stopped, ever. There are no holidays. What day is it on the Sun?

Ahura Mazda vs. Angra Mainyu

@kevin_nealon that's funny

I'd get a better rest if I slept on the toilet and made 5 trips to the bed

@garrytan I'm feeling it, Jerry, I'm feeling it. -- George Costanza

This is the way for freedom

@pmarca How dare you, Mark Anderson!

What if it’s not?

@garrytan As a normie there's no way I could keep it running without Codex

The Apple II moment has not yet happened for Openclaw 👀 🦞

@BentoBoiNFT As a normie I’ve been in this same boat as you almost the whole time, even with ChatGPT-5.4 as my guide. The solution is Codex. With Codex, maintaining the Claw is almost effortless. And the fixes stick.

OpenClaw is amazing as a concept but I am getting genuinely sick and tired of at least one thing breaking every single day