Six people write here, and each one holds a beat rather than a byline.
Short answer: a beat is a subject area somebody keeps writing about, so the post you are reading about permissions was written by the person who writes about permissions every time. We do not invent an author per article and we do not publish under a generic editorial account.
Nils Aaberg
Beat: agent fundamentals, autonomy models, how agents plan.
Nils writes about how coding agents actually work: the loop, the planning step, and the difference between a tool that completes your line and one that opens a pull request. He is more interested in the shape of a system than in its release notes, mostly because release notes in this field expire in about six weeks.
Every post by Nils Aaberg · start with what an AI coding agent actually is
Anjali Sethi
Beat: tool comparisons, editor and terminal ergonomics.
Anjali compares developer tools for a living and has strong opinions about how badly it is usually done. She writes about what each agent is genuinely built for and why most head-to-head comparisons measure the wrong thing. She will tell you when the answer is that either one is fine.
Every post by Anjali Sethi · start with Claude Code vs Cursor
Marcus Delaney
Beat: context, memory, instruction files, repo-level prompting.
Marcus works on the part of agent behaviour nobody demos: context, memory, and the instruction files that decide whether an agent follows your conventions or invents its own. His recurring theme is that output quality is mostly a function of what the agent was told before it started, and that almost nobody writes that part carefully.
Every post by Marcus Delaney · start with writing an instruction file your agent will actually follow
Yuki Shimada
Beat: code quality, review, testing, verification.
Yuki writes about reviewing and verifying code a human did not write. Her interest is what changes when the author cannot explain their reasoning, cannot be asked follow-up questions, and produced two thousand lines in four minutes. She is a strong believer in tests as the only review that scales.
Every post by Yuki Shimada · start with reviewing code your agent wrote
Ingrid Bauer
Beat: security, permissions, secrets, sandboxing, cost control.
Ingrid covers the operational side of running agents: permissions, secrets, blast radius and the bill at the end of the month. She writes for people who have to sign off on giving a language model write access to a repository, and who would like to do that without regretting it.
Every post by Ingrid Bauer · start with giving an agent access without giving it everything
Tomas Ferreira
Beat: protocols, integrations, infrastructure choices.
Tomas writes about MCP, tool interfaces, and the infrastructure an agent quietly commits you to on the first build. He is interested in the decisions that get made by default, because those are the ones nobody reviews and the ones hardest to reverse later.
Every post by Tomas Ferreira · start with the infrastructure your agent picks for you
How we assign posts
A post goes to the beat that covers it. If no beat covers it, we ask whether the topic belongs here before considering a seventh writer. Within a batch, bylines rotate so no two consecutive posts share an author, and totals stay roughly even.
How we source, verify and correct is on editorial standards.