compositional-writing

Atomic writing treats each paragraph or card as a single, self-contained, reusable unit to minimize cognitive load and token costs when producing code comments, docs, logs, and prompts.

Make documentation grep-friendly by using stable IDs, structured frontmatter, and surface metadata so content can be easily indexed, searched, and navigated at scale across large knowledge bases.

A Map of Content (MOC) structure connects atomic writing cards via tags and indexes, directing readers from a central SKILL.md file to precise references for easy navigation and scalable discovery.

Structure technical articles using a six-field frame model encompassing what, why, when, where, how, and acceptance criteria to align intent with implementation and guide surface-driven reading.

Yes, compositional writing enforces a multi-pass review workflow that applies iterative reviews to steadily improve clarity, correctness, and layer alignment in atomic, grep-friendly content.

Use atomic writing for code comments and logs when you need to minimize token costs, reduce cognitive load, and ensure each content unit is self-contained, reusable, and easily discoverable via grep.