How to Use Agent Skills - The AI Daily Brief: Artificial Intelligence News and Analysis Recap

Podcast: The AI Daily Brief: Artificial Intelligence News and Analysis

Published: 2026-03-18

Duration: 28 min

What Happened

Anthropic's new Dispatch feature for Claude Cowork has quickly gained traction with 3 million views and 9,000 bookmarks on its announcement tweet. Dispatch allows users to manage and monitor Claude sessions from their mobile devices, offering flexibility for tasks like creating reports or booking flights while on the go. Users such as Daniel Sahn and Ethan Mollock have positively reviewed the feature, noting its stability and integration with tools like Gmail.

Interest in OpenClaw is rising in China, despite recent government warnings against its use. Concerns about privacy and security have prompted Chinese regulators to issue guidelines for deploying OpenClaw, highlighting the cultural resonance of new technology in China. Analysts like Wendy Chang and Tony Pang have noted a shift in sentiment from excitement to anxiety, driven by fears of job displacement.

Meta's acquisition of Manus has caught the attention of Chinese authorities, who are scrutinizing the deal over concerns about data exports and talent loss. The relocation of Manus' headquarters to Singapore and its U.S. venture backing have sparked regulatory interest. Jensen Huang of Meta remains optimistic about securing export approvals and continuing operations in China.

Amazon CEO Andy Jassy has predicted that AI could double AWS revenue to $600 billion annually. This ambitious forecast suggests a new growth phase for cloud services, driven by AI's impact on the total addressable market. Analysts like Patrick Moorhead see this as a repricing of cloud services, with companies like Amazon poised to benefit significantly.

The episode also discusses Anthropic's development of agent skills, designed to streamline AI capabilities without overwhelming context windows. Skills are structured as directories with markdown files, scripts, and resources that agents can utilize dynamically. This modular approach helps agents load specific knowledge as needed, improving efficiency and reliability.

Tariq from the Claude Code team shares insights on building effective skills, emphasizing the importance of progressive disclosure and maintaining a gotcha section. The Skill Creator tool is highlighted as a way to test and benchmark skills, ensuring they trigger correctly and remain relevant with model updates. Anthropic's approach to skills is gaining traction across various AI ecosystems, including Notion's recent introduction of custom skills for Notion AI.

Key Insights