When you spawn an agent to write a chapter or critique a draft, the context you pass determines whether it stays on canon or invents contradictions. This teaches you when to use file flags versus session history, what each agent type actually needs (writers want style files and continuity anchors, not the whole manuscript), and when to materialize ephemeral decisions before handoff. The judgment calls matter more than the mechanics. Pass a brainstormer too much context and it plays it safe instead of exploring. Skip the vocab file and your writer picks the wrong faction name. Most useful when orchestrating multi-agent story pipelines where each spawn needs different scope.
npx -y skills add haowjy/creative-writing-skills --skill story-context --agent claude-codeInstalls into .claude/skills of the current project.
Select a file.
supercent-io/skills-template
supercent-io/skills-template
huangjia2019/claude-code-engineering
reactjs/react.dev
reactjs/react.dev