Bad Agent covers Anthropic, Karpathy, Slack

This reads less like a launch and more like a sharp field report on where AI developer conversations are actually converging: security, agent ops, and collaboration UX.

• –Karpathy joining Anthropic keeps the pretraining talent war and frontier-lab competition front and center
• –The GitHub Actions and VS Code extension discussion is a reminder that AI-era attack surfaces keep widening beyond model APIs
• –The Slack angle is the interesting product thesis: coordinating multiple agents may be a messaging problem before it is a model problem
• –Best for builders who want a fast synthesis of current agent, infra, and security discourse rather than a formal product deep dive