What problem does it solve? Writing Claude Code output styles incorrectly causes lost coding behavior, broken frontmatter, and confusion between styles, skills, and artifacts. This Skill guides the creation, review, and plugin packaging of output-styles Markdown files so they behave safely and predictably. ## Core Features & Use Cases - Frontmatter Authoring: Produces correct name, description, keep-coding-instructions, and plugin-only force-for-plugin fields with a machine-readable schema. - Boundary Routing: Decides whether a request belongs in an output style, a skill, CLAUDE.md, an agent, or an artifact, preventing mixed surfaces. - Plugin Packaging & Safety Review: Covers plugin output-styles/ directories, manifest outputStyles overrides, forced-style risks, and coding-instruction preservation. - Use Case: A user asks for a project-level style that leads every coding response with an architecture decision and risk table; the Skill produces .claude/output-styles/architecture-brief.md with keep-coding-instructions: true and a system-prompt body. ## Quick Start Ask the assistant to create a Claude Code output style for your desired tone and format, specifying the target scope and whether coding instructions should be preserved.