Bento Deck Authoring

Problem

bento-slides targets the task of turning source material into a directly openable presentation file. The input may be data, process steps, specs, images, or media; the output is a self-contained .bento.html. It is not intended for plain pasted paragraphs, because the skill expects content to be mapped into structured presentation features such as charts, tables, morph transitions, state slides, ken-burns motion, and media.

Workflow and Caveats

The workflow starts by locating or fetching an empty shell, then reading the JSON inside #bento-doc. New documents should read the schema first, keep required fields such as size and theme, and omit docId and collab; existing decks must not change docId. Element fields need to be fully specified, such as stroke/strokeWidth for shapes and fontFamily/align/valign for text, while respecting 1280×720, 96px side margins, one accent color, and at most two typefaces. Before delivery, the skill self-audits whether numbers should be charts, whether consecutive slides on one subject share id and transition:"morph", whether there is at least one motion moment, whether a drill-down should become a state slide, and whether speaker notes exist on every slide. Media can embed short clips as data URIs, while large files should use URLs, and autoplaying video must be muted. Local file:// preview may be blocked by browser extensions, so a temporary HTTP server can be used for verification.