Virtual companies of role-playing agents keep showing up. What the research says about them, and what a background team of agents looks like when it drops the theatre.

Long agent sessions get worse, and ending them loses things. What a Claude Code skill does at the end of each session, and where the same ideas show up elsewhere.

boxed-ui, a small Claude Code mod that puts your prompts and the agent's replies in their own frames in the terminal.

When an agent explains a system to us, it often draws a diagram, and until recently that diagram was Mermaid. Mermaid takes a list of boxes and arrows and decides the layout itself. That works until the layout carries the point: the database belongs to the right of the app,

An English translation of DeepSeek's article on Super-agent architecture: a resident router, isolated worker sessions, event-driven triggers and an external watchdog.

How a human and an AI agent keep shared context alive between sessions, and how the Open Knowledge Format gives that context a shape.