OpenClaw intel for builders β signal, not hype.
OpenClaw is an open-source AI agent platform that can autonomously operate your computer β browsing the web, running apps, managing files, and executing multi-step workflows on your behalf. It was originally released as "Clawdbot," later rebranded to Moltbot, and is now known as OpenClaw.
OpenClaw can be installed via npm (npm install -g openclaw), Docker, or by cloning the GitHub repo. See our full installation guide for step-by-step instructions on every platform.
Yes. OpenClaw is fully open-source under the Apache 2.0 license. You can use, modify, and distribute it freely. Some cloud-hosted versions or premium ClawHub skills may have separate pricing.
OpenClaw grants broad system permissions by design, so caution is warranted. The project has integrated VirusTotal scanning for ClawHub skills, but security researchers continue to find malicious packages. Always review a skill's source, limit permissions, and keep OpenClaw updated. See our troubleshooting guide for security tips.
Skills are plugin-like extensions distributed via ClawHub that add new capabilities to your OpenClaw agent β like browsing the web, managing cloud infrastructure, or interacting with APIs. Think of them like npm packages but for agent actions.
In February 2026, OpenClaw creator Peter Steinberger joined OpenAI. OpenClaw remains open-source under a community foundation, and OpenAI has committed to keeping it that way. Read our timeline for details.
OpenClaw competes with platforms like BitBuddies, Adept, and Moltbot/Emergent. Its key differentiators are the open-source model, the ClawHub skill ecosystem, and broad OS-level control. See our alternatives comparison.
Check out our ELI5 explainer for a beginner-friendly overview, our usage & tutorials page for hands-on guides, and our glossary for key terms.
The latest release delivers a redesigned macOS app settings UI, new QA parity and toolβcoverage checks, faster Gateway restarts, and broader reliability fixes across Telegram, Discord, Codex/OpenAI, and xAI OAuth. It also adds plugin tooling for simple tool plugins, new skills, and various performance and stability improvements. (github.com)
- Googles New AI Is The OpenClaw Killer β AI Revolutions β May 16, 2026 β `https://www.dailymotion.com/video/xa8dh0i` β A fast-paced news commentary comparing Googleβs rumored βRemyβ agent to OpenClaw and what that rivalry means for agentic AI. (dailymotion.com)
Anthropic reversed an earlier ban by introducing capped βAgent SDKβ credits so Claude subscribers can again power thirdβparty agents like OpenClaw, but programmatic usage is now metered and separated from interactive chat quotas. The change restores compatibility while curbing runaway token costs. (venturebeat.com)
- OpenClaw + Ollama + Kimi K2.5 is INSANE (FREE!) β Julian Goldie SEO β May 12, 2026 (estimated; upload date not shown) β `https://vimeo.com/1161705906` β A practical tutorial showing how to run OpenClaw entirely with local models via Ollama and Kimi K2.5 to avoid API costs. (vimeo.com)
TechRadarβs Q&A highlights growing SMB adoption and Hostingerβs oneβclick managed OpenClaw offering, alongside practical security guidance like limiting permissions, isolating deployments, and monitoring agent actions. It frames OpenClaw as an alwaysβon assistant that automates routine tasks across familiar tools. (techradar.com)
GLAD documents a TOCTOU issue in OpenClawβs OpenShell fileβwrite path affecting versions before 2026.4.22 and recommends upgrading to 2026.4.22 or later; references include the NVD and the associated GitHub security advisory. (advisories.gitlab.com)
This release continues externalizing cloud providers and plugins to slim core dependencies, adds a modelsβauth listing command, and folds in wideβranging stability and auth fixes across Discord, Gemini, Codex, diagnostics, and more. It advances the shift toward safer defaults and clearer operational tooling. (github.com)
- 2026β05β11 β Autonomous Agents: A Show and Tell of OpenClaw (Recording) β Drupal AI Learners Club (YouTube) β May 11, 2026 β `https://www.youtube.com/watch?v=3dGyjqO_PUY` β A live demo session that walks through OpenClawβs setup and shows it scaffolding a Drupal project while discussing realβworld security and cost considerations. (drupal.org)
Stack.watch tracks several newly indexed OpenClaw issues (e.g., SSRF, auth bypasses, environmentβvariable injection, and tokenβreuse flaws) affecting versions prior to lateβApril fixes, underscoring the need to run 2026.4.22+ and audit plugin and sandbox configurations. The page provides plainβEnglish synopses and CVE IDs for quick triage. (stack.watch)
- ClawBox β Your Own AI Assistant Running 24/7 | NVIDIA Jetson β Π―Π½ΠΊΠΎ ΠΠ»Π΅ΠΊΡΠ°Π½Π΄ΡΠΎΠ² β May 10, 2026 (estimated; upload date not shown) β `https://vimeo.com/1164052005` β A concise hardware demo of a β¬399 Jetson Orin Nano box running OpenClaw with voice control, browser automation, and personal workflow skills. (vimeo.com)
- Using Personal AI in 2026 (OpenClaw) β Anna D β May 7, 2026 (estimated; upload date not shown) β `https://vimeo.com/1188033410` β A businessβoriented talk that reviews what OpenClaw is, how to deploy it quickly, and practical workflows it can automate today. (vimeo.com)
- OpenClaw, Obsidian & Hermes β PJ Glasso β May 6, 2026 (estimated; upload date not shown) β `https://vimeo.com/1185031583` β A short demo of a local βAI command centerβ stack that combines OpenClaw orchestration with Hermes messaging and Obsidian knowledge management. (vimeo.com)
OpenClaw is an open-source AI agent platform that can autonomously operate your computer β browsing the web, running apps, managing files, and executing multi-step workflows on your behalf. Originally released under a different name, OpenClaw has rapidly become one of the most talked-about projects in the AI-agent space thanks to its extensible "skill" system (distributed via ClawHub) and its ability to chain actions across local and cloud environments.
Peter Steinberger releases Clawdbot, a personal AI-agent experiment that can control a desktop computer autonomously. The project garners early attention from the hacker community.
Word-of-mouth spreads fast. The project rebrands to Moltbot as download counts surge. Developers start building third-party "skills" β plugin-like extensions that chain agent actions.
A companion product, Moltbook, is teased for notebook-style agent workflows. Soon after, the entire project rebrands again to OpenClaw, emphasizing its open-source ethos and the new ClawHub skill marketplace.
Critical vulnerabilities surface β including CVE-2026-25253 (one-click RCE via Control UI) β prompting rapid patches and VirusTotal integration for ClawHub skills. Security researchers begin auditing the ecosystem extensively.
Reports emerge that multiple Big Tech companies, including OpenAI and Google, are in discussions about acquiring or integrating OpenClaw. The AI-agent space heats up as competitors race to match OpenClaw's capabilities.
OpenClaw creator Peter Steinberger joins OpenAI. Crucially, OpenClaw remains open-source under a community foundation β OpenAI commits to supporting, not acquiring, the project. The Verge, Financial Times, and Business Insider all cover the story.
A news hub that aggregates the latest OpenClaw updates, security advisories, release notes, and community chatter. We focus on signal over hype.
No. This is an independent publication. We are not affiliated with, endorsed by, or sponsored by the OpenClaw project or its maintainers.
ClawHub is OpenClaw's community marketplace for "skills" β plugin-like extensions that add capabilities to the agent. Think of it like a package registry (npm, PyPI) but for agent actions.
OpenClaw grants broad system permissions by design. The project has integrated VirusTotal scanning for ClawHub skills, but security researchers continue to find malicious packages. Always review a skill's source, limit permissions, and keep OpenClaw updated.
Content is refreshed periodically based on news flow. Check the "Last updated" timestamp at the top of the news section.
A critical remote-code-execution bug in OpenClaw's Control UI that allowed one-click token exfiltration via a malicious link. It was patched in v2026.1.29 (January 30, 2026). All users should update immediately.
Not yet β we're a static v0 site. Future versions may accept community submissions. For now, all content is manually curated and reviewed before each update.
We aggregate recent coverage from major tech and security outlets, then compile it into a structured feed.