This week saw major moves in AI-powered coding tools and security concerns. Mistral AI launched an Agents API that lets multiple AI helpers work together on complex coding tasks, like fixing bugs or searching the web. New Relic teamed up with GitHub Copilot to automatically spot and fix code problems, showing how AI agents are joining the DevOps world.

Security researchers warned about slopsquatting - when AI tools invent fake code package names that hackers can use to sneak in malware. A new study found this happens 20% of the time. JetBrains announced a $12,000 contest with Mistral AI to build better code completion tools that understand programmers' context.

Developers debated pricing for AI helpers like Claude, with some finding $200/month too steep despite good performance. These changes show AI becoming both more powerful in coding workflows and more risky if not used carefully.