Skills & Tools

Google ADK Python v1.32.0 introduces significant enhancements to agent development capabilities, including native OpenTelemetry metrics, Anthropic thinking blocks support, improved authentication mechanisms (2LO/3LO via GcpAuthProvider), and express mode deployment. The release addresses critical security vulnerabilities (RCE via YAML, SSRF), fixes parallel tool execution issues, and improves analytics with BigQuery LLM cache metadata logging and credential isolation to prevent race conditions.

Claude Code v2.1.126 introduces gateway model listing, project purge functionality, improved OAuth authentication for WSL/SSH/containers, and numerous bug fixes. Key improvements include permission bypass options, better shell detection on Windows, security fixes for managed settings, and stability enhancements for streaming, image handling, and remote sessions. The release also refines telemetry events, fixes UI issues, and resolves edge cases in tool availability and clipboard handling.

crewAI v1.14.4 introduces enhanced persistence customization, expanded cloud provider support (Azure OpenAI, Vertex AI), and new research tools (Tavily, You.com MCP). The release includes 12 bug fixes addressing JSON handling, message preservation, credential forwarding, and multimodal input support. Documentation updates cover E2B and Daytona sandbox tools.

Claude Code v2.1.121 introduces significant usability and performance improvements, including a new `alwaysLoad` MCP option for tool availability, enhanced plugin management with pruning, and improved terminal UI with scrollable dialogs and better URL handling. The release addresses critical memory leaks affecting image processing and session history, fixes OAuth and authentication issues across multiple platforms, and adds support for Workload Identity Federation and custom redirect URIs. Multiple terminal rendering bugs are resolved, and performance optimizations reduce file descriptor usage and startup time.

This guide demonstrates how to build Claude Managed Agents with persistent memory capabilities that learn and retain user preferences across multiple sessions. Users create a memory store, configure a shopping agent with file access tools, and show how the agent automatically reads and updates customer preferences on repeat visits. The memory system uses a simple file-based approach mounted at /mnt/memory/ that both the agent and application can access via API.

OpenClaw v2026.4.25-beta.2 delivers major upgrades across text-to-speech (TTS) with new providers and chat-scoped controls, plugin architecture optimization through cold persisted registry, comprehensive OpenTelemetry observability expansion, safer browser automation with CDP improvements, and hardened installation/update processes across all platforms. The release includes PWA/Web Push support in the control UI, per-agent and per-account TTS overrides, and deterministic plugin discovery that eliminates broad manifest scans.

OpenClaw v2026.4.25-beta.3 introduces major upgrades to text-to-speech (TTS) with new providers and chat-scoped controls, migrates plugin management to a cold persisted registry for faster startup, expands OpenTelemetry observability across model calls and tool loops, enhances browser automation safety with iframe-aware snapshots, and hardens installation/update processes across multiple platforms. The release also adds PWA/Web Push support to the control UI and includes new CLI tools for setup and diagnostics.

OpenClaw v2026.4.25-beta.4 introduces major upgrades to text-to-speech (TTS) with new providers and chat-scoped controls, migrates plugin startup paths to a cold persisted registry for faster performance, expands OpenTelemetry observability across model calls and tool loops, hardens browser automation with safer URLs and iframe awareness, and strengthens installation/update processes across multiple platforms. The release also adds PWA/Web Push support to the control UI and includes new CLI tools for plugin management and diagnostics.

OpenClaw v2026.4.24 introduces Google Meet as a bundled participant plugin with real-time voice capabilities, adds DeepSeek V4 models to the default catalog, and enhances browser automation with coordinate clicks and improved action budgets. The release significantly lightens plugin and model infrastructure at startup through static catalogs and lazy dependencies, while implementing breaking changes to plugin SDK tool-result transforms that require migration to the new `registerAgentToolResultMiddleware` API.

OpenClaw v2026.4.24-beta.1 introduces Google Meet as a bundled participant plugin with real-time voice capabilities, adds DeepSeek V4 models to the default catalog, and implements WebRTC-backed realtime voice loops across Talk, Voice Call, and Google Meet. The release includes browser automation improvements (coordinate clicks, longer action budgets, per-profile headless overrides), lighter plugin/model infrastructure at startup, and breaking changes to plugin SDK tool-result transforms requiring migration to `registerAgentToolResultMiddleware`.

OpenClaw v2026.4.23-beta.6 introduces significant enhancements to image generation across multiple providers (OpenAI, OpenRouter), adds optional forked context for subagents, implements per-call timeout support for generation tools, and includes configurable local embedding context sizes. The release also delivers numerous fixes for Codex harness routing, media handling, provider authentication, and UI persistence, improving reliability across chat platforms including Slack, Telegram, WhatsApp, and WebChat.

OpenClaw v2026.4.23 introduces significant enhancements to image generation across multiple providers (OpenAI, OpenRouter), adds optional forked context for subagents, implements per-call timeout support for generation tools, and includes configurable local embedding context sizes. The release also fixes numerous issues related to Codex harness routing, media handling, provider authentication, and streaming behavior across multiple platforms including WhatsApp, Slack, and Telegram.

OpenClaw v2026.4.23-beta.4 introduces significant enhancements to image generation across multiple providers (OpenAI, OpenRouter), adds optional forked context for subagents, implements per-call timeout support for generation tools, and includes configurable local embedding context sizes. The release also delivers numerous fixes for Codex harness routing, media handling across platforms (WhatsApp, Telegram, Slack), transcript replay behavior, and provider authentication flows to improve reliability and user experience.

OpenClaw v2026.4.23-beta.5 introduces significant enhancements to image generation across multiple providers (OpenAI, OpenRouter), adds optional forked context for subagents, implements per-call timeout support for generation tools, and includes configurable local embedding context sizes. The release fixes numerous issues across Codex harness routing, media handling, provider authentication, and UI persistence, while updating Pi dependencies to v0.70.0 and improving debug logging for harness selection decisions.

OpenAI Agents Python v0.14.5 release includes three key fixes and features: Modal sandbox idle timeout configuration, HITL resume tool output serving, and streamed terminal output backfilling. Documentation improvements clarify lazy skill source host paths. This patch release addresses sandbox management, human-in-the-loop workflows, and output streaming reliability.

OpenClaw v2026.4.20-beta.2 is a major release featuring UI/UX improvements to the onboarding wizard, strengthened agent prompts, tiered model pricing support, session maintenance optimizations, and numerous provider-specific enhancements for OpenAI, Anthropic, Moonshot Kimi, and other integrations. The release includes 40+ fixes addressing security, transport normalization, thinking mode handling, and plugin dependency management.

OpenClaw v2026.4.20 is a major release featuring UI/UX improvements to the onboarding wizard, enhanced agent prompts with better completion bias and verification, support for tiered model pricing including Moonshot Kimi models, session maintenance optimizations to prevent memory issues, and numerous fixes across plugins, APIs, and provider integrations. Key improvements include streaming thinking in Mattermost, better group-specific system prompts in BlueBubbles, and refined reasoning model handling across OpenAI GPT-5 and other providers.

OpenAI Agents Python v0.14.3 is a maintenance release focused on Windows compatibility and stability improvements. Key fixes include normalizing sandbox paths for Windows support, making sandboxes importable on Windows, preparing Daytona workspace roots, and improving manifest handling. The release also adds Datadog as an external tracer option and includes documentation updates.

OpenAI Agents Python v0.14.2 release introduces sandbox extra path grants, MongoDB session backend support, and tool origin metadata persistence. The release includes 11 bug fixes addressing sandbox parsing, tool removal, response handling, and session management. Documentation updates and dependency bumps round out the release with contributions from 9 new community members.

langchain-anthropic version 1.4.1 is a maintenance and feature release that includes support for Claude Opus 4.7 features and adaptive thinking mode, fixes for null encrypted_content in compaction blocks, dependency updates (langsmith 0.7.31, pytest 9.0.3), and security patches for Pygments CVE-2026-4539. The release also updates model profiles and improves documentation with typo fixes.