buildnotes is a skill for coding agents that turns the history of a software feature into an explanation of how and why it was built. It reads issues, pull requests, commits and reviews, then selects decisions to organize into chapters. The resulting explanation can include figures drawn from the real product, screenshots and code. Its homepage presents the tool for Claude Code, Codex and OpenCode, with requests made inside the repository being explained.
The output format is part of the request. A post produces a page with chapters and figures that can be published directly or moved to another site. A video presents the same story as a self-playing piece, while a deck turns it into slides and a PDF. Individual figures can also be requested as an image, GIF, video or SVG. Published examples include a room-walkthrough feature, a before-and-after screenshot workflow and a weekly release digest. These examples show the intended presentation style rather than establish that the tool has verified every claim shown inside a demonstration.
The homepage names Node 22, git and Chrome as prerequisites and links to its own GitHub installation instructions, scripts and visual-component documentation. The product is useful for developers who have a substantial implementation record but have not turned that record into a readable explanation. Generated material should still be checked against the actual changes, particularly when it describes private work, reasons for a decision or a release promise. Access to a repository does not itself make every issue, comment or screenshot appropriate for publication. The central workflow is explanation from implementation evidence, with an editor deciding what can be shared and whether the resulting narrative matches the work.
