Kunlun Tech Launches 'SkyAgents' AI Agent Development Platform


AI Safety Center's Remote Labor Index reveals a breakthrough in remote work automation. Claude Fable5 achieved a record 16.1% automation rate on freelance tasks like 3D modeling, architectural design, graphic design, and video animation, based on industry standards measuring AI output quality matching or surpassing human professionals.....
Apple WebKit team introduced MCP server in Safari Technology Preview 247, using AI agents to simplify front-end development and debugging. MCP (Model Context Protocol) is an open standard that enables AI agents to connect to external tools, databases, local files, and browser dev tools, allowing information access and authorized actions, bridging the data gap between AI and dev environments, improving programming agent collaboration efficiency.....
Perplexity jointly released a report with the Harvard Business School, comparing the Perplexity Computer general AI agent with traditional search assistants. Traditional assistants only answer questions and require users to perform follow-up actions manually, while AI agents can autonomously plan, execute tasks, and produce results. Data shows that Perplexity Computer's AI agent runs autonomously for an average of 26 minutes per session, far exceeding traditional search assistants, demonstrating the comprehensive transformation of knowledge work by AI agents.
The WeChat AI ecosystem has undergone a major upgrade. On June 8, 2026, it will open up AI Agent capabilities for developers, allowing them to access AI services within mini programs. Meituan, as one of the first internal test partners, has completed the development and testing together. In the future, users will be able to directly use Meituan takeout and other local life services through the WeChat Agent, achieving an intelligent lifestyle experience across scenarios and with high complexity, driving the intelligent transformation of local life services.
The open-source AI agent framework MateClaw releases version v1.5.0, focusing on optimizing the underlying infrastructure for team operations, including three key features: target acceptance, knowledge base consistency maintenance, and multi-user memory isolation. It addresses the issue of traditional agents relying solely on ambiguous completion score assessments for tasks, enhancing management transparency.