How to use the Novel Writing Toolbox Router agent skill (story)

What is an agent skill and what does story do?

An agent skill is a folder with a SKILL.md file that tells an AI agent when and how to perform a task. The story skill is the routing layer: it reads your request and sends it to the matching helper, such as long-form planning, short-story drafting, analysing a published book, reviewing a draft for consistency, or cover prompts.

The routing table in the source maps intents to specific skills. For example, requests about planning a long novel go to long-write, requests about checking a draft for consistency go to review, and requests about a cover go to cover. If your request does not match clearly, the skill asks only for the missing detail, such as whether the novel is long or short.

Install the story skill

Download the ZIP from this page and unzip it into one of these locations: ~/.claude/skills/ for your user account, or .claude/skills/ inside a project folder for that project only.

Alternatively, in the Claude apps go to Settings → Capabilities → Skills and upload the ZIP there.

• Personal install: unzip into ~/.claude/skills/
• Project install: unzip into the project's .claude/skills/ folder
• Claude apps: Settings → Capabilities → Skills, then upload the ZIP

Trigger the skill with natural requests

You can call the skill with its command, or simply describe what you want. The source lists these triggers: /story, $story, /story dashboard, /网文, and phrases such as 我想写小说 (I want to write a novel), 记住我的写作习惯 (remember my writing habits), 打开工作台 (open the workspace), and 检查更新 (check for updates).

Calling style differs by host. Claude Code and OpenCode use /story-* commands, while Codex and ZCode use $story-*. If you just type /story without a clear intent, the skill shows four plain options: open or continue a long novel, write a short story, make a passage less AI-sounding, or see more advanced options.

• Plan a long novel: 讨论长篇结构、规划剧情、写大纲 (discuss structure, plan plot, write an outline)
• Analyse a book: 拆文、分析这本书 (break down a text, analyse this book)
• Check a draft: 审查、一致性检查 (review, consistency check)
• Reduce AI tone: 去 AI 味 (remove the AI flavour)
• Open the dashboard: /story dashboard

Launch the local dashboard

When you run /story dashboard, or ask to open the workspace or view project files, the skill starts a local Dashboard. It uses the current working directory as the workspace unless you name another directory, which must already exist. The skill runs scripts/dashboard-server.mjs with Node.js and prints a local URL for you to open.

The dashboard can browse and edit .md, .txt, .json, .yaml, .yml and .toml files. When saving or deleting, it checks file content versions to avoid overwriting external changes. Stop it by ending the Node process.

Check for updates before upgrading

Ask whether a newer version exists and the skill reports the installed version against the latest release. It tries the GitHub CLI first, then curl. It reports the current status and does not install anything until you agree.

If the lookup fails, the skill tells you it could not fetch the latest version and points you to the Releases page rather than showing an error.

• Uses gh release view if the GitHub CLI is available
• Falls back to curl against the GitHub releases API
• Updates only after you say yes

Safety notes: scripts and network access

This skill runs scripts and uses the network. The dashboard server listens only on 127.0.0.1 by default. The skill is instructed not to add --allow-network or expose the workspace to a local network or the internet.

The dashboard serves local project files to any browser on the same machine through the loopback address. If your manuscript is sensitive and the computer is shared, do not run it there.

• Dashboard binds to 127.0.0.1 by default
• Do not use --allow-network on shared machines
• Update checks contact api.github.com or the GitHub CLI