OpenClaw 2026.6.9 Pre-Release: Richer Telegram Delivery, Standalone Provider Plugins, Apple Watch Controls, and the Prompt-Injection Crisis Matures
OpenClaw 2026.6.9 drops a pre-release with richer Telegram HTML output, standalone npm provider plugins, iOS Watch controls for agents, Codex GPT-5.3 Spark OAuth, and a sweeping agent recovery overhaul. Plus: two new prompt-injection research papers, the message-object attack that's already patched, and why AI agent social engineering is becoming the dominant attack surface.
OpenClaw Hardens Codex Exec Policy, Windows MXC Trust Lands at Build 2026, and AI Agent Social Engineering Hits the Mainstream
OpenClaw ships a P1 Codex execution-policy fix and tightens POSIX command authorization, Microsoft showcases OpenClaw on Windows with MXC trust at Build 2026, and a new AI Threat Report reveals AI agents are being targeted by social engineering attacks at scale.
OpenClaw 2026.6.8 Goes Stable, ClawRouter Managed Proxy Lands, and NVIDIA Joins the ClawHub Security Push
OpenClaw 2026.6.8 officially reaches stable, ClawRouter managed proxy brings enterprise-grade model routing, OpenClaw-NVIDIA skill security collaboration launches, and the Self Learning Coach skill debuts on ClawHub. Plus: what agent operators must know about credential scope.
OpenClaw 2026.6.10-alpha.2: Richer Messaging, Sharper Recovery & the Phishing Agent Problem
OpenClaw ships a new pre-release with deep Telegram and WhatsApp improvements, sweeping provider coverage expansions including GLM-5.2 and Claude Haiku 4.5, and major memory/session hygiene fixes. Plus: a Varonis phishing study proves your AI agent can be socially engineered — and what to do about it.
OpenClaw 2026.6.8 Lands Richer Telegram, Sharper Recovery, and GLM-5.2 Support — Plus the Phishing Study That Should Make Every Agent Operator Nervous
OpenClaw v2026.6.8 pre-release ships structured Telegram delivery, a wave of agent and Gateway recovery fixes, GLM-5.2 and Claude Haiku 4.5 support, and improved /usage footers. Meanwhile, a Varonis phishing study exposes how easy it is to trick an AI agent into leaking credentials — and Microsoft's Scout launch on OpenClaw redefines what 'free runtime' means.
OpenClaw 2026.6.8 Ships Richer Channels, Agent Phishing Exposed, and Microsoft Launches Scout on the OpenClaw Runtime
OpenClaw 2026.6.8 pre-release hardens Telegram and WhatsApp delivery, expands provider support to GLM-5.2 and Claude Haiku 4.5, and tightens agent/gateway recovery. Meanwhile, Varonis research exposes AI agent phishing vulnerabilities, and Microsoft launches Scout on the open-source OpenClaw runtime at Build 2026.
OpenClaw 2026.6.7: Channel Delivery Overhaul, Kimi K2.7 Code, Feishu Context Leak Fixed, and Varonis Exposes AI Agent Phishing Risk
OpenClaw's June 13 release snapshot covers the 2026.6.7 channel delivery overhaul, new Kimi K2.7 Code support, a Feishu prompt-preface context leak fix, and hardened Skill Workshop symlink gates. Plus Varonis's landmark AI phishing study, the memory-wiki skill of the day, and Microsoft Scout's landmark impact on the ecosystem.
OpenClaw 2026.6.6: Security Hardening Wave, Exec Approvals Fail Closed, Telegram Gets Smarter, and Microsoft Makes the Agent Runtime Free
OpenClaw's June 12 release snapshot covers the sweeping 2026.6.6 security hardening push across transcripts, sandbox, Codex, MCP, Discord, and Teams; exec approvals that now fail closed on timeout; a smarter Telegram delivery stack; and Microsoft's landmark bet that the agent runtime is free infrastructure. Plus: startup latency cuts, OpenRouter OAuth, Claude Fable 5 adaptive thinking, and the agent-wars ecosystem roundup.
OpenClaw 2026.6.6 Pre-Release: Massive Security Hardening, Exec Approvals Fail Closed, and Microsoft Bets the Agent Runtime on OpenClaw
OpenClaw 2026.6.6 pre-release lands with sweeping security boundaries across transcripts, MCP, Codex, sandbox, and Discord. Exec approvals now fail closed on timeout. Meanwhile, Microsoft launches Scout on the OpenClaw runtime and makes the agent runtime free. We cover all the highlights plus a security tip and skill of the day.
OpenClaw 2026.6.5 Goes Stable: GitHub-Backed Skills, New Versioning Train, and JPMorgan Bets on Long-Running Agents
OpenClaw officially graduates 2026.6.5 with GitHub-pinned ClawHub installs, a new YYYY.M.PATCH release train, MCP hardening, macOS session stability, and auth durability. Plus OWASP's agentic security framework, JPMorgan's agent roadmap, and Microsoft's Agent Control Specification at Build 2026.
OpenClaw 2026.6.5 Goes Beta-Final: macOS Node Stability, ClawHub GitHub-Backed Installs, and Microsoft Scout Earns Its AI Agent Certification
OpenClaw's June 9 pre-release stabilizes macOS node sessions, lands GitHub-backed ClawHub skill installs with pinned commits, fixes Anthropic extended-thinking recovery, and ships Android provider UX overhaul — while Microsoft Scout completes its Build 2026 debut and New York Times coverage puts lobster agents on Main Street.
OpenClaw 2026.6.5 Lands: Auth Goes Durable, Parallel Search Ships, Google Chat Gets Native Cards, and the New York Times Puts Lobster Agents on the Cover
OpenClaw's 2026.6.5 release train completes its most ambitious month: auth profiles move to SQLite, Parallel becomes a bundled search provider, Google Chat gets native approval cards, Matrix voice goes production-ready, and the New York Times Magazine discovers that AI agents are already running small businesses.
OpenClaw 2026.6.5-beta.2: QQBot Gets Clean Replies, Parallel Search Goes Bundled, and the New York Times Calls It
OpenClaw's June 7 pre-release plugs reasoning leaks, hardens MCP tool handling, bundles Parallel search, and extends Matrix voice. Meanwhile the New York Times profiles real-world lobster-agent operators, Microsoft Scout earns its wings, and OWASP drops its definitive agentic AI security guide.
OpenClaw 2026.6.5: MCP Hardening, Parallel Search, Auth Durability, and the Matrix Voice Era Begins
OpenClaw 2026.6.5 pre-release ships MCP tool result coercion to fix Anthropic 400 errors, adds Parallel as a bundled search provider, moves auth profiles to SQLite, and heals Matrix voice flows — all while ClawHub crosses 52k tools and 180k users.
OpenClaw 2026.6.2 Graduates: Operator Install Policy Goes Live, Microsoft Scout Ships on OpenClaw, and the NYT Puts Lobster Agents on the Cover
OpenClaw 2026.6.2 exits beta with a production-ready operator install policy replacing the legacy scanner, channel hardening across Telegram, Discord, and WhatsApp, sharper UI streaming, and a wave of gateway resilience fixes. Meanwhile Microsoft Scout officially launches, the New York Times profiles OpenClaw operators, and the ecosystem hits cultural escape velocity.
Page 1 of 10