FreedomIntelligence’s OpenClaw Medical Skills library brings 869 agent skills to the biomedical space — covering clinical workflows, genomics, drug development, bioinformatics pipelines, and protein design. The most specialized collection in the ClawHub ecosystem.
Two projects are pushing the OpenClaw ecosystem to its hardware minimum: MimiClaw runs a full AI agent on a $5 ESP32-S3 chip in bare-metal C, with no OS and 0.5W power draw. PycoClaw takes a MicroPython approach with more features and broader hardware support. Here’s how they compare.
Most OpenClaw agents are brilliant at reasoning, writing, and working with local files — but ask one what happened in the news an hour ago and you’ll hit a wall. That’s the gap Felo AI set out to close. Their recently launched suite of eight interconnected ClawHub skills turns any OpenClaw agent into a real-time
On March 3rd, 2026, OpenClaw became the most-starred software project on GitHub, passing React’s 243,000 stars in roughly 60 days. React needed a decade. Here’s the timeline, the context, and what it actually means.
Most AI agents talk to the world through Slack or Discord. That works — until you need end-to-end encryption, self-hosted infrastructure, or the ability to federate across organizations without handing your data to a third party. The @openclaw/matrix plugin, which shipped as part of OpenClaw v2026.3.22, brings the Matrix protocol into the OpenClaw ecosystem as
What if your AI agent could look at its own mistakes, figure out what went wrong, and fix itself — without you lifting a finger? That is the promise of Capability Evolver, a meta-skill on ClawHub that has quietly become one of the most downloaded skills in the registry. With 27,700+ downloads and nearly 500
Google Research’s TurboQuant compresses LLM key-value cache by 6x with zero accuracy loss. OpenClaw already supports local backends like Ollama and LM Studio — but breakthroughs like this are what make fully local, fully capable agent setups genuinely practical.
From a rural Austrian village to 145,000 GitHub stars and a call from Sam Altman: the story of how Peter Steinberger burned out building a billion-device PDF company, disappeared to Madrid, came back to vibe-code an AI agent in one hour, and accidentally built the most talked-about open-source project of 2026.
One of the biggest limitations of AI agents is their ephemeral memory. Once a session ends, context vanishes. But a growing category of ClawHub skills is tackling this head-on, giving OpenClaw agents persistent memory, structured knowledge bases, and the ability to learn from past interactions. If you’ve ever wished your agent could remember what you
OpenClaw’s biggest update of the year landed last week. ClawHub is now the default skill registry, sub-agents got multi-model support and adjustable thinking levels, session bloat is finally managed automatically, and a new /btw command lets you ask quick questions without polluting your main context.