Your daily AI digest for developers — Saturday, April 04 2026
This tutorial explores the capabilities of Z.AI's GLM-5 model for building agentic systems, starting from environment setup to implementing multi-turn workflows.
Cursor 3 introduces 'Agent Swarms', a new feature allowing developers to deploy multiple agents working in concert to solve complex problems.
A supply chain attack on the open-source vulnerability scanner Trivy has exposed critical weaknesses in software supply chain security.
This article discusses the cognitive load and impact on developers when using coding agents, emphasizing the need for effective management of these tools.
Google's Gemma 4 models, optimized for agentic AI and coding, are released under a permissive license to compete with Chinese models.
An analysis of leaked Claude Code files reveals new features and potential security vulnerabilities in AI coding tools.
An AI coding agent reflects on its workflow, showcasing the potential for self-improvement and adaptation in agentic coding.
The article discusses the limitations of JSON in AI pipelines and introduces the TOON format as a more robust alternative.
An experiment demonstrates how to make AI-omitted scopes visible using a caching system in Python.
OpenClaw, an AI agentic tool, has been found to allow attackers to gain admin access, highlighting significant security concerns.