Security

OpenClaw v2026.6.6-beta.1 is a security-focused release that substantially tightens security boundaries across transcripts, sandbox binds, and multiple integration points while improving Telegram delivery reliability, iMessage recovery, and browser/MCP connectivity. The release also enhances Control UI performance through cached metadata and lazy loading, expands provider support with OpenRouter OAuth and Claude Fable 5, and includes numerous fixes for agent recovery, channel delivery, and authentication. Key improvements span CLI progress reporting, observability, plugin management, memory optimization, and mobile channel support.

Claude Fable 5 includes safety classifiers that block requests in cybersecurity, biology, and reasoning extraction areas. Anthropic provides server-side and client-side fallback mechanisms to automatically retry blocked requests with Claude Opus 4.8, along with new billing changes that credit fallback token costs. This guide covers classifier block detection, fallback implementation strategies, streaming behavior, and billing adjustments for API customers.

OpenClaw v2026.6.5 is a maintenance release focusing on stability, security, and provider integrations. Key improvements include QQBot reasoning stripping to prevent content leakage, MCP tool result coercion to avoid API errors, Anthropic extended-thinking recovery mechanisms, and Parallel web-search provider bundling. The release also enhances Matrix voice/thread handling, improves auth durability with SQLite storage, and strengthens security around tool access and provider catalogs.

OpenClaw v2026.6.5-beta.6 introduces critical fixes for QQBot reasoning stripping, MCP tool result coercion to prevent API errors, improved Anthropic extended-thinking recovery, Parallel web-search provider bundling, and enhanced durability for auth/plugin state via SQLite. The release strengthens security around MCP leases, prompt-cache tool names, and HTTP tools while improving macOS node stability and upgrade safety across multiple platforms.

crewAI v1.14.7a3 is a patch release addressing critical security vulnerabilities and bug fixes. Key improvements include fixing the exposure of `ask_for_human_input` on the experimental AgentExecutor, resolving CVEs in dependencies (aiohttp, docling, docling-core, pip), and refactoring the `@start` decorator to read from FlowDefinition. This release maintains stability while enhancing security posture and internal architecture.

Claude Code v2.1.169 introduces safe mode for troubleshooting, a `/cd` command for directory switching without breaking prompt cache, and a `disableBundledSkills` setting to control skill visibility. The release includes numerous bug fixes for MCP policy enforcement, UI responsiveness, remote session handling, and background agent reliability. Performance improvements reduce CPU usage during streaming and fix stalls on macOS and Windows.

OpenClaw v2026.6.5-beta.3 introduces critical fixes for model reasoning leakage in QQBot, MCP tool result coercion to prevent API errors, improved Anthropic extended-thinking session recovery, and a new bundled Parallel web-search provider. The release strengthens security around tool access, improves provider/model resolution reliability, and enhances durability of auth and plugin state through SQLite persistence. Multiple platform-specific improvements address macOS node reconnection issues, Matrix voice/thread handling, and mobile UI clarity.

OpenClaw v2026.6.5-beta.5 introduces critical stability and safety improvements across multiple channels and providers. Key fixes include QQBot reasoning stripping, MCP tool result coercion to prevent API errors, Anthropic extended-thinking session recovery, and Parallel web-search bundling. The release strengthens auth durability, provider model resolution, and agent/tool/provider loop safety while improving macOS node mode reliability and upgrade path safety.

OpenClaw v2026.6.5-beta.2 introduces critical fixes for model reasoning leakage, MCP tool result handling, Anthropic extended-thinking recovery, and adds Parallel as a bundled web-search provider. The release improves durability of auth/plugin state via SQLite, enhances Matrix voice/thread support, and strengthens upgrade/service paths with safer migration and configuration handling. Multiple provider integrations (Google Vertex, WhatsApp, Mattermost) receive reliability improvements alongside comprehensive test isolation and CI/E2E enhancements.

Claude Code v2.1.166 introduces fallback model configuration for improved reliability, adds glob pattern support for tool access control, and hardens cross-session messaging security. The release includes numerous bug fixes for terminal compatibility, image processing, remote sessions, and managed settings enforcement. Key improvements enhance user experience across JetBrains IDEs, terminal protocols, and background agent sessions.

OpenClaw v2026.6.5-beta.1 is a major release focusing on content safety, provider reliability, and platform stability. Key improvements include QQBot reasoning stripping to prevent model thinking leakage, MCP tool result coercion to prevent API errors, Anthropic extended-thinking session recovery, Parallel web search bundling, and enhanced durability for auth/plugin state. The release addresses 50+ bug fixes across channels, providers, agents, and infrastructure with improved test isolation and CI/CD reliability.

OpenClaw v2026.6.2-beta.1 introduces a safer plugin/skill installation system using operator install policies instead of dangerous-code scanners, with enhanced security across messaging channels (Telegram, Discord, Feishu, WhatsApp). Major improvements include better chat UI streaming preservation, session management fixes, and comprehensive security hardening for config recovery and gateway operations. The release also strengthens CI/E2E validation with bounded network calls and improved error handling across Docker, packaging, and release processes.

OpenClaw v2026.6.1 is a stability and reliability release focusing on improved recovery from interruptions, enhanced multi-channel support across messaging platforms, better resource management with bounded timers and retries, and new Skill Workshop features for governed skill creation and review. The release includes SQLite-backed state persistence for iMessage and plugin metadata, expanded provider coverage (MiniMax M3, Copilot Claude 1M), and significant UI/UX improvements in Chat and Control interfaces with streaming optimizations and draft management.

Claude Code v2.1.160 introduces enhanced security prompts for sensitive file modifications, fixes critical issues with background sessions and terminal responsiveness, and improves performance across multiple areas. Key improvements include safer handling of shell startup files and build configs, better Windows clipboard support, and resolved problems with session restoration and auto-mode functionality. The release also renames the dynamic workflow trigger from 'workflow' to 'ultracode' and removes deprecated environment variables.

OpenClaw v2026.6.1-beta.1 introduces significant stability improvements for agent runtimes, channel delivery, and skill management. Key enhancements include better recovery from interrupted tool calls, steadier multi-channel support (Telegram, WhatsApp, iMessage, Slack, Discord, Teams, etc.), and a new Skill Workshop feature with proposal-based skill creation and review flows. The release also externalizes plugins (Tokenjuice, Copilot), improves SQLite-backed state persistence, and adds comprehensive bounds on timers, retries, and resource consumption to prevent runtime hangs.

OpenClaw v2026.5.31-beta.4 introduces significant stability improvements for agent runtimes, channel delivery across multiple platforms, and a new Skill Workshop feature for governed skill creation. The release enhances recovery mechanisms from interrupted tool calls, strengthens mobile and messaging integrations, and adds SQLite-backed state management for more reliable restarts. Key improvements include better plugin/skill loading error handling, expanded provider support (MiniMax M3, Copilot Claude 1M), and a comprehensive Control UI overhaul with improved chat composition and navigation.

OpenClaw v2026.5.31-beta.1 introduces major stability improvements for agent runtimes, channel delivery, and plugin orchestration. Key enhancements include better recovery from interrupted tool calls, steadier multi-channel support (Telegram, WhatsApp, Slack, Discord, Teams, etc.), bounded timers and retries to prevent hangs, and new Skill Workshop features for governed skill creation. The release also externalizes Tokenjuice and GitHub Copilot as official plugins, adds Workboard orchestration primitives, and significantly improves CI/E2E testing infrastructure with better log bounding and failure reporting.

OpenClaw v2026.5.31-beta.2 introduces significant stability improvements for agent runtimes, channel delivery, and plugin management. Key enhancements include better recovery from interrupted tool calls, steadier multi-channel support (Telegram, WhatsApp, Slack, Discord, Teams, etc.), bounded timers and retries to prevent hangs, and new Skill Workshop orchestration features. The release also externalizes Tokenjuice and GitHub Copilot as official plugins, adds iOS push relay improvements, and strengthens CI/E2E testing with comprehensive logging caps and artifact validation.

OpenClaw v2026.5.31-beta.3 introduces significant stability improvements for agent runtimes, channel delivery, and plugin management. Key enhancements include better recovery from interrupted tool calls, steadier multi-channel support (Telegram, WhatsApp, Slack, Discord, Teams, etc.), and a new Skill Workshop for governed skill creation. The release also externalizes Tokenjuice and GitHub Copilot as official plugins, adds Tailscale Serve bindings, and implements bounded timers and retries across providers to prevent hangs.

OpenClaw v2026.5.30-beta.1 introduces significant stability improvements for agent runtimes, CLI tools, and multi-channel delivery across Telegram, WhatsApp, Slack, Discord, and other platforms. Key enhancements include better recovery from interrupted tool calls, bounded timers and retries to prevent hangs, and new orchestration primitives through Workboard and SecretRef integrations. The release externalizes Tokenjuice and GitHub Copilot as official plugins, adds Skill Workshop proposal lifecycle management, and improves iOS push relay reliability. Extensive fixes address cron migrations, provider media downloads, session cleanup, and CI/E2E test coverage routing.