Posts

If you’ve been running OpenClaw for more than a few weeks, Anthropic’s newly launched Agent Skills will feel instantly familiar — and that’s not a coincidence. Released this week on the Claude platform, Agent Skills are modular capability packages that extend Claude’s functionality across conversations. Each Skill bundles instructions, metadata, and optional resources — scripts, templates, workflows — that Claude loads automatically when relevant to a user’s request. Sound familiar? It should. ...

The escalating tension between Anthropic and the open-source agentic AI ecosystem hit a new flashpoint this week when Anthropic temporarily suspended the account of OpenClaw creator Peter Steinberger — the same week the company announced it was no longer covering OpenClaw usage under flat-rate Claude subscriptions. What Happened Early Friday morning, Steinberger posted to X with a screenshot of a message from Anthropic informing him his account had been suspended for “suspicious” activity. His accompanying caption was blunt: “Yeah folks, it’s gonna be harder in the future to ensure OpenClaw still works with Anthropic models.” ...

One of the persistent friction points in agentic AI commerce has been a deceptively simple problem: you don’t always know how much a task will cost before you start it. Token-intensive LLM inference, variable-complexity queries, and dynamically-scoped compute jobs all have costs that only crystallize at completion. The x402 protocol — Coinbase’s open HTTP payment standard for AI agents — has until now required fixed, exact prices upfront. That made it great for deterministic APIs, but it structurally excluded an entire tier of usage-based services. ...

At RSAC 2026, four separate keynotes from four separate companies arrived at the same conclusion without coordinating: zero trust must extend to AI agents. Microsoft, Cisco, CrowdStrike, and Splunk each named AI governance as the biggest gap in enterprise security. The problem, as Cisco’s Matt Caulfield put it, isn’t just authenticating an agent once and letting it run — it’s that “at any moment, that agent can go rogue.” Now two vendors have shipped architectures that actually try to solve it. Anthropic and Nvidia have each published zero-trust AI agent frameworks — and they solve the credential isolation problem in fundamentally opposing ways. ...

One of the hardest problems in agentic AI development isn’t building a single capable agent — it’s designing systems where multiple agents coordinate effectively without breaking down. Anthropic has stepped into that gap with a comprehensive published framework covering five multi-agent coordination patterns, giving developers an authoritative reference for the architectural decisions that matter most when building complex autonomous systems. Why Coordination Patterns Matter A single agent can handle a bounded task well. But real-world complexity quickly exceeds what any single agent can do in a single pass — tasks require verification, parallelization, specialization, and feedback. The naive approach (just make one very smart agent) breaks down at scale: context windows fill up, error rates compound, and there’s no mechanism for catching mistakes before they propagate. ...

Anthropic’s Claude Cowork has officially moved from preview to general availability — and it arrived with a serious enterprise upgrade. Starting today, all paid plan users on macOS and Windows can access Cowork with new capabilities that push squarely into territory previously dominated by specialized RPA and desktop automation tools. The headline feature is Dispatch — and it’s exactly as bold as the name implies. What Is Dispatch? Dispatch is a new Claude Cowork capability that lets Claude directly control macOS applications using mouse, keyboard, and screenshots. Think of it as Claude gaining hands: it can see your screen, click buttons, type into forms, and navigate GUIs the same way a human would — without requiring API integrations or custom automation scripts. ...

For years, the “agent economy” has been a thought experiment. Researchers and futurists asked: what happens when AI agents can spend money, sign contracts, and act on behalf of businesses without human intervention at every step? Today, Meow Technologies made that question significantly more concrete. The San Francisco fintech announced the launch of what it describes as the first agentic banking platform — a financial infrastructure layer designed not for humans, but for AI agents acting as autonomous business actors. ...

If you’re running OpenClaw and haven’t updated past version 2026.3.25, stop what you’re doing and patch now. Two high-severity CVEs dropped today — on the same day — targeting OpenClaw’s gateway authentication layer and session management. Both are confirmed, both are actionable, and both are the kind of vulnerabilities that keep security engineers awake at night. CVE-2026-35669: Scope Boundary Bypass (CVSS 8.8) The first and more severe vulnerability, CVE-2026-35669, was published April 10, 2026 by TheHackerWire, carrying a CVSS score of 8.8 (High). That’s a significant rating — anything above 7.0 is considered high-severity, and 8.8 puts this squarely in the zone that demands immediate attention. ...

Anthropic dropped something significant yesterday: a full research framework titled Trustworthy Agents in Practice, published alongside the launch of Claude Managed Agents. It’s the clearest public articulation yet of how Anthropic thinks about safe, autonomous AI agent deployment — and it directly addresses the two biggest failure modes the industry is grappling with right now. Why This Matters Now AI agents are no longer prototype toys. Claude Code, Claude Cowork, and a growing ecosystem of third-party deployments are completing multi-step tasks — writing and running code, managing files, browsing the web, interacting with APIs — with minimal human supervision. That autonomy is the whole point. It’s also precisely where things can go wrong. ...

Atlassian is betting that the biggest problem with enterprise knowledge management isn’t creating content — it’s transforming it. Today’s announcements for Confluence put two new capabilities directly at the format bottleneck: AI Remix for visual transformation of existing pages, and MCP-powered partner agents that push Confluence content directly into Lovable, Replit, and Gamma. Both are in open beta now. The partner agents start rolling out to Confluence Cloud customers on April 13. ...