@tibo_maker The Style just reminds me too much of early 2000s ringtone commercials.
Ringedingdingding
English
Matthias Kampmann
387 posts
Matthias Kampmann
@M_Kampmann
B2B Revenue Operations, Project Management & Data Integration | Remote Working Dad | Metalhead
Germany Katılım Mayıs 2013
211 Takip Edilen87 Takipçiler
@itsolelehmann Seems long?
I use a short version that is directionally similar:
Do not take shortcuts. Do research before implementing and start with red-green TDD. Build the whole thing right: tests, docs, and a permanent fix. No workarounds. No "later."
English
Garry Tan’s custom instructions are based.
It pushes any LLM past half-finished bullshit and into actually useful answers.
Add this to your SOUL md:
English
Man I‘m coding this space goblin Game and Codex is super bad at it…
English
@_lhermann Not sure the 1M Context window will be enough for all that red tape though
English
I added a German Law tool to my home-brew OpenClaw alternative. Now it can look up any regulation and help me win against German paperwork ⚔️
English
@powerbottomdad1 Only if it validates my point.
If not I use the work experience Joker
English
my entire dev pipeline is basically just repeating a series of 6 skills 20 times a day...
/research
/build
/but-for-real
/review
/learnings
/pr
English
@petergyang Exact reason I did Not touch Hermes yet.
I could spend a week playing with it or get someactual work done with the current openclaw setup
English
The thing I have to remind myself constantly with OpenClaw, Claude Code, and the like is am I actually setting up workflows to get real work done or am I just optimizing my OpenClaw and Claude Code setup.
Very easy to just do the latter because it feels like progress (and is fun) but isn't actually super productive after a certain extent.
English
When you get married you think you're just getting a wife.
But you're also getting 3 more pillows added to your bed every year of mariage.
English
@onlinedopamine GLM 5.1 as orchestrator and handling all the Cron stuff.
If it touches code or infra it delegates to a Codex subagent.
English
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
English
@rileybrown I fondly remember Turbo Pascal and Basic from my school days.
Ancient and almost mythical
Sit down by the fire…
English
@askobylarz Simple rule in our house: No spending Limit on books
English
Always interesting to me when I recommend a book and people ask if its worth it. It's a book, at most $10 - $20. Just buy it, even one takeaway will pay dividends.
English
@svpino I wish this were true.
Tool calling is not reliable though.
Also it plays back the instructions of repeating jobs instead of executing them quite often.
GLM 5.1 though…
English
If you ask me, Gemma 4 is one of the best models out there for a single reason:
You can run it locally and it’s really, really good (probably Sonnet level?)
And, of course, ya can now use it to power OpenClaw and show a middle finger to the company who doesn’t like it.
atomic.chat@atomic_chat_hq
Run OpenClaw with Gemma 4 and Atomic Chat MacBook Air M4 · 16 GB RAM · 25 tok/s No cloud! No subscription fees! Open-source local model. Runs on your regular device
English
@thekitze Same when using Codex vanilla.
But I dont get that vibe when I use an Agent as orchestrator that spawns Codex Sessions.
Only rarely the „did you do the thing?“ - „lel I was just pretending brb“ pattern
English
my experience with gpt 5.4 in openclaw
- pls continue
- yes pls do that
- yes do what i told you to do
- yes continue
- please do what you said you're gonna do
- did you do that? no. then pls do it.
just pure pain
English
@_lhermann Tried to go with API credits but ripping through them like crazy.
I’m now on GLM 5.1 as Orchestrator and handing coding to Codex.
English
Claude Subscription no longer working with OpenClaw is a big blow. I had so many cool things automated with it.
How are you guys handling this? I'm switching to cron tasks that run claude code directly in a docker container. It was easier than I thought actually.
English
Since Anthropic blocked third party harnesses (openclaw) I've been testing open source models as orchestrators for my agents.
A month ago that led to sub par results.
Now with qwen 3.6 and glm 5.1 you can barely tell the difference.
English
@itsolelehmann Provider lock in and if you Need to share the kb across multiple agents
If that’s not a real issue for you I’d also not add the complexity
English
why would I use obsidian when I can just use claude code for the knowledge base?
whats the advantage?
English
@itsolelehmann Doing the Same here. Also much easier if you need to Check in via mobile alot.
Tailscale + tmux are perfect
English
I might just go for the setup of
1. Mac Studio at my parents house to run local models (because we will be traveling most of the year wo a permanent home rn) + tailscale
2. buy a lighter macbook pro
seems like the better setup
Ole Lehmann@itsolelehmann
is it worth it to get the m5max macbook pro? I want to be able to run local models. how much ram should I get? can't get a mac studio because I'm traveling a lot this year.
English
@michael_chomsky Immediate B2B VC funding if the lines Match the customer’s CI
English
@trq212 Throw some bread crumbs at the unwashed masses?
Classy
English
claim a month of free credits on us, thanks for bearing with us
Boris Cherny@bcherny
Subscribers get a one-time credit equal to your monthly plan cost. If you need more, you can now buy discounted usage bundles. To request a full refund, look for a link in your email tomorrow. support.claude.com/en/articles/13…
English
