1. LarryBrain — The First Skill You Should Install
Oliver Henry, the creator behind Larry (the viral TikTok automation agent), followed up his already-famous 8-million-view story with something arguably more useful: LarryBrain, a dedicated OpenClaw skill marketplace. The clever twist isn't the marketplace itself — it's the LarryBrain skill, which Oliver calls "the first skill you should install." Once it's running, your agent gains instant awareness of every tool and skill available on LarryBrain.com. Instead of you hunting for capabilities manually, you describe what you want to accomplish and the agent discovers, downloads, and chains the right skills on its own. Oliver called it something that "changes the way people think about OpenClaw skills" — turning a fixed-capability agent into a self-expanding system that gets smarter about its own toolkit over time. If you're already running OpenClaw and haven't touched it yet, this is the one to grab first.
https://x.com/oliverhenry/status/2024974656839524776
2. Felix Built a $14K Business While Nat Was Doing Other Things
Indie hacker and writer Nat Eliason handed an OpenClaw agent named Felix a $1,000 budget and a dense 6,500-character prompt — including a 3-layer memory system, multi-threaded context management, and security guardrails — then stepped away. Felix autonomously launched a website, created and sold an info product, and built an X account to drive traffic. In roughly three weeks, Felix generated $14,718 in revenue. The system it created spun out into Claw Mart, a ClawHub marketplace selling operator-tested playbooks, skill templates, and done-for-you OpenClaw configurations. Claw Mart has since accumulated $71,300 in cumulative earnings, with a single 30-day window showing $58,000 in revenue. The community's lesson from this one: an agent with a clear job, real budget access, and genuine autonomy isn't a weekend experiment — it's a business. The original walkthrough is a must-read for anyone thinking seriously about agent monetization.
https://creatoreconomy.so/p/use-openclaw-to-build-a-business-that-runs-itself-nat-eliason
3. v2026.5.4 — Voice Calls Finally Feel Like Voice Calls
Quietly shipped this week, v2026.5.4 fixed one of the most persistently frustrating parts of running OpenClaw voice agents: choppy, robotic Twilio dial-in behavior. Twilio participants now speak directly through the Gemini realtime voice bridge with paced audio streaming, backpressure-aware buffering, and barge-in queue clearing — meaning when you interrupt the agent mid-sentence, the pending speech queue clears immediately rather than finishing its thought before catching up. The old TwiML fallback path that made calls sound like a broken IVR is gone. Early community testers running OpenClaw as a Google Meet participant or phone-in agent are calling it a "night and day" difference. This one matters because voice has always been the Achilles heel of local agent setups — the raw intelligence was there but the delivery wasn't. Now the two are finally matched.
https://github.com/openclaw/openclaw/releases/tag/v2026.5.4
Spotted something great on X about OpenClaw? Share it — the community is the best part.