How Linear works with coding agents: what we learned

On August twentieth, Linear's coding agent started attaching before and after screenshots to prove its own work. We build proof for coding agents, so we studied how Linear works: its docs, its changelog and essays, and the threads where users explain why they stay. Work lands in triage, where one key accepts, declines or snoozes. Accepted work moves into cycles, projects and releases, and agents are assigned like people at every step. People love six things. It is fast, it lives on the keyboard, it has opinions, you can feel the care, the code moves the issue, and agents are teammates. Open a pull request and the issue moves to in progress. Merge it, and it is done. And when an agent takes the issue, you stay accountable. Linear's agent attaches screenshots for the reviewer. Vidmatic builds for the person who asked: the ask is a recording, and a narrated video of the fix lands in their Inbox. Against all of this, we found seven gaps. We file into GitHub only, there is no team board, merges do not move our cards, and there is no command menu. Our Inbox cannot snooze, we ship without a public changelog, and our pages wait for the server. Cycles and estimates stay with Linear, which does them well. So Vidmatic should be the proof, posted wherever the work lives, with statuses nobody has to type. The full list is on the Vidmatic blog.