SUB30
SUB30
Loading...
Tips, guides, and news from the ClawHosters team.
On March 20, Anthropic launched Claude Code Channels. The feature connects Claude Code sessions to Telegram, Discord, and iMessage through MCP-based plugins. You message Claude from your phone, it processes the request using your local dev environment, and replies in the same chat.
Jensen Huang stood on stage at GTC 2026 and told Jim Cramer something that sent shockwaves through the AI industry. CNBC captured it live: "This is definitely the next ChatGPT." He called OpenClaw "the most popular open-source project in the history of humanity." Not Linux. Not React. OpenClaw....
OpenClaw v2026.3.22 shipped on March 23, 2026. It packs 242 commits from 70+ contributors, and if you run a ClawHosters instance, you probably care about two things most: your agents can now run for up to 48 hours, and the plugin ecosystem just got a proper SDK.
This one is a hotfix. OpenClaw v2026.3.22 shipped with packaging bugs that broke the dashboard UI (503 errors) and left the WhatsApp standalone plugin out of the npm bundle entirely. v2026.3.23 fixes that, plus 49 other issues across authentication, messaging, and LLM configuration. 16...
Your OpenClaw instance is running. Users are chatting with it. But do you actually know what's happening inside?
If you starred an OpenClaw repo on GitHub recently, check your notifications. Scammers are tagging developers in fake issues promising $5,000 in "CLAW tokens." There is no CLAW token. There never was.
OpenClaw's `system.run` allowlist got broken three different ways in March 2026. The allowlist is supposed to block unapproved commands from running on your machine. It didn't. Three separate vulnerabilities, three different bypass techniques, one patch release.
A team tweaked one system prompt. Looked fine in testing. In production, average reasoning chains jumped from 4 steps to 11. Token spend doubled. No error logs, no alerts, no HTTP 500s. They found out when the invoice arrived. That story, documented by Agentix Labs, is probably the best argument...
The company launched ADP Claw (Agent Development Platform), a full deployment pipeline that lets enterprise admins push OpenClaw agents directly into WeChat, WeCom, and QQ. Three configuration steps. That's it. Your OpenClaw agent shows up in work group chats, ready to handle queries from...
March 21, 2026 was a rough day for OpenClaw self-hosters. A coordinated disclosure dropped over 10 high-severity CVEs hitting core OpenClaw components. That's the largest single-day CVE batch in the project's history.
Most AI agents are smart but uninformed. They know the internet. They don't know your company's return policy, your internal API docs, or what your team decided in last Tuesday's meeting. That's the gap retrieval augmented generation fills.
CrowdStrike's CTO Elia Zaitsev just published what might be the most thorough security breakdown of OpenClaw to date. The title alone says a lot: "What Security Teams Need to Know About OpenClaw, the AI Super Agent."
