What problem does it solve? Measuring multiline text in the browser normally requires DOM reads like getBoundingClientRect, which cause reflow thrash and make animated text layouts impractical. This Skill uses the @chenglou/pretext library to compute line breaks, per-line widths, and per-grapheme positions purely via canvas measurement, enabling text effects that CSS cannot produce. ## Core Features & Use Cases - Variable-width text flow: Stream lines around moving obstacles (sprites, ASCII logos, wire shapes) using layoutNextLineRange with per-row width functions at 60fps. - Text-as-geometry demos: Build games where bricks or obstacles are measured words, shatter typography into per-glyph particles, and drive kinetic type with per-line transforms. - Multiline shrink-wrap and measurement: Compute exact box heights and tightest container widths without touching the DOM, for virtualized lists and masonry layouts. - Use Case: A user asks for a hero section where a paragraph parts around a draggable glowing orb. The Skill starts from the hello-orb-flow.html template, swaps in real prose, tunes the palette and font, and delivers a single self-contained HTML file. ## Quick Start Ask the agent to build a pretext demo where text flows around a moving shape, and it will generate a self-contained HTML file you can open with python3 -m http.server.