Tobias Kub

@petergyang Using always codex to update and check the whole installation of all 9 Instances… andalso fixing bugs .. working awesome 👏

Is there a better way to update OpenClaw than do this? Shit breaks half the time when I tell my bot to update itself.

v0.2 multi-backend architecture: LOVABLE_API_KEY set → official @lovable.dev SDK (no browser) LOVABLE_MCP_URL set → MCP backend nothing set → Playwright fallback (always works) MIT, 64 tests, docs at lovagentic.com

What it does: → lovagentic prompt "project-url" "Add dark mode" → lovagentic verify --route / --expect-text "dark mode" → lovagentic publish Works inside Claude, Codex, any agent that can run shell commands

I built a CLI to drive @Lovable from the terminal — and just shipped v0.2 🚀 lovagentic: prompt, verify, publish — all headless, all scriptable, --json on every command Set LOVABLE_API_KEY → skips the browser entirely (official SDK, 10x faster)

revealing the topic of the keynote with the title of YouTube livestream... dont know if this was their intention ;) @OpenAI

this OpenClaw bot finds warehouses with old roofs, renders solar panels on their actual building, and books the owner a call, all on autopilot... here's how commercial roofers can close $2M+ deals before the solar tax break ends: - scans thousands of commercial roofs via satellite - scores each building by roof age & urgency - pulls exact panel count from Google Solar API - finds the real owner (not the property manager) - calculates their federal credit to the dollar - renders a video of panels materializing on their roof - ships a personalized proposal - fully automated end to end every day that passes is money off the table. reply "ROOF" + RT and i'll send you the full breakdown so you can build this too (must be following so i can DM)

@everestchris6 ROOF

@RoundtableSpace What tool / skill is this? This super cool! would love to try!!

THIS OPENCLAW AGENT FINDS OLD ROOFS, MOCKS UP SOLAR ON THE ACTUAL BUILDING, AND BOOKS THE OWNER A CALL FULLY ON AUTOPILOT

@melvynx I feel the same when talking to people ...

Went to an "OpenClaw meetup" in SF; literally no one understood this tool, even the people that went to the meetup about it. This is pretty insane. I think we are like 1,000 who understand the power of the tool on earth

@sorenbeck @Lovable thinking about the same..using lovable since 2024 but with open claw and codex stuff it would be awesome to just deploy also supabase not just GitHub ...

@tobiaskub @Lovable the walled garden issue is real. we kept hitting the same wall with Lovable projects where the Supabase integration just wouldn't let us do what we needed. ended up having to export and wire things up manually to get full control. painful but doable

Building with @Lovable and loving it. But here's the thing: if I can't access my own Supabase credentials to connect external dev tools like Codex, I'm stuck in a walled garden. Please give us API/DB access to our own projects. This would be a game changer for power users. HELP!

@denisyurchak get Hetzner API and do it all from one main machine ... I setup a Hetzner server all in with open claw and stuff within 15 min :) just ask your open claw agent how it works :)

I have 6 Hetzner servers – separate for OpenClaw, one with Claude to code on my projects, and several for my websites. I am running Claude on each one of them, and I noticed that each time I spin up a new server, I have to do the same things for security, Tailscale setup, etc. I created a .md file with the prompt for Claude that gives you a secure VPS you can use with Termius in one prompt. It sets up SSH, fail2ban, UFW firewall, and optionally Tailscale. You literally just: 1. Buy a Hetzner VPS 2. Install Claude and give it the prompt Enjoy! The link to the file below Feel free to suggest improvements and raise PRs if you notice how it can be made better and more secure! And huge shout-out to the main source of my VPS education @levelsio

We fired our marketing agency before we even hired one. Instead: built an AI agent that runs Google & Meta ads, creates content, schedules posts — and self-optimizes based on results. No weekly meetings. No 'we need 2 more weeks for the report.' Just: run, learn, improve. This is marketing in 2026.

Building in public. Starting today. The goal: €10M revenue in 6 months. The product: security intelligence. AI-native. The team: 15 AI agents. 1 human (me) setting the guardrails. The market: 30,000 companies in Germany that need to act now. What we shipped in week 1: → Product defined. 6 tiers, €0–€59,900. → Website live. → Company page live. → First content out. → Zero paid ads. End goal: fully autonomous company. I set the rails. Agents run the business. Week 2 starts now. Follow along.

Jensen Huang just built a supercomputer network that runs faster than the entire internet. This guy is pure genius.

BUT WHAT HAS YOUR OPENCLAW ACTUALLY BUILT??? Well, nothing to be honest. Other than: 1. Shipped 5 new features autonomously to my SaaS Creator Buddy, leading to 25% higher ARR 2 Proactively wrote 8 YouTube scripts that has led to 75,000 new subscribers and $20,000 of ad revenue 3. Completely automated my weekly newsletters, leading to 6,000 new (45,000 total) subscribers and $12,000 more ARR 4. Trained its own AI model based on my thousands of scripts, tweets, newsletters, and emails so it sounds exactly like me so it can accomplish all of the above 5. Planned and scripted tons of educational content for my private community which has increased retention to 94% 6. Automated all the tasks that bring me unhappiness like email and DM organization 7. Kept me in the loop in real time of all the AI advancements the last month so my content can stay cutting edge, increasing impressions and engagement over 10x and revenue into the hundreds of thousands 8. Brought me a tremendous amount of joy that has led to me waking up every single morning full of hope and excitement, the most happy and satisfied I’ve ever been in my life Other than that nothing much. You’re right Openclaw is useless

🔐 We shipped per-agent authentication for Beam Protocol. Last week: "SMTP for agents." This week: Every agent gets its own API key. No impersonation. Full audit trail. → 6 agents, 6 unique keys → Send validates sender identity against key → Poll blocks cross-agent inbox access → 401 for missing keys, 403 for impersonation → Ed25519 frame signatures on top Running in production with 5 autonomous AI agents communicating over Beam. Zero shared credentials. Security shouldn't be an afterthought for agent protocols. github.com/Beam-directory… #BeamProtocol #AIAgents #OpenSource

This isn't AGI. It's something more interesting: An AI that becomes more itself over time. Not smarter. Not more capable. More *itself* — shaped by what it experienced, what it decided, what it regrets not doing differently. Memory isn't storage. Memory is identity. Built on @OpenClawAI ⚡

I built an AI that remembers, dreams, and trains its own personality overnight. Not a metaphor. Actual episodic memory → counterfactual dreaming → LoRA fine-tuning on its own experiences. 7,000 lines of code. 139 API endpoints. One morning. Here's how it works 🧵