story-review Skill: Multi-Reviewer Checks for Your Web Novel Draft

Reviews your web novel draft from several angles at once and returns a plain report of what must change, what to improve and what can stay.

Installs: 16.7K; on skills.sh
Stars: 8.2K; on GitHub
Files: 21; besides SKILL.md
Updated: Nov 2026; last repo commit

Run /story-review, or add lean or solo. Full mode starts four helper agents (story architect, character designer, narrative writer, consistency checker) in parallel. Lean uses two. If agents are missing or fail, it falls back to a single-session review.

It runs bundled scripts first: a punctuation check and an AI-pattern check on the chapter files. Platform rubrics cover Fanqie, Qidian and Zhihu. On long novels it can also update tracking files. Full mode needs /story-setup in your project, and the skill is written in Chinese.

Good for

Chinese web novel authors who draft with a coding agent
Writers working on Qidian, Fanqie, Jinjiang, Qimao or Zhihu Yanxuan
Authors who want a second opinion before posting a chapter

Install story-review

Claude Code

Copy this prompt.

Install the story-review skill in Claude Code. Run this command:

npx skills add zenstory-ai/oh-story-claudecode -y -g

Then check that it loaded and tell me how to use it.

2
Paste it into Claude Code. It shows the change and asks before running it.

Requires Node.js.

This installs the whole oh-story set of 14 skills, as the README says. Then open your writing folder, run /story-setup, and start a new session so the helper agents load. Drop -g to install into the current folder only.

From the project README. Source.

What's in the skill

When it loads

多视角对抗式审查。full/lean 模式在已部署 reviewer agents 时并行 spawn；缺失/异常 agents 或 spawn 失败时自动降级 solo，参考文件不可读时使用内置 rubric fallback。触发方式：/story-review、/审查、「审查一下」「帮我审一下」。

Sections in SKILL.md

作者偏好边界
Review Mode 选择
Phase 0：预检与降级（必须先执行）
审查基准与参考资料规则（必须遵守）
报告面向作者（必须遵守）
参考资料解析顺序
内置审查基准包（路径不可读时必用）
Phase 1：收集待审查内容
统一 Findings Schema（所有模式必须使用）
Phase 2：并行 Spawn Agent（full/lean 模式）
Phase 3：综合裁决（full / lean 模式）
Phase 4：输出报告（full / lean 模式）
solo 模式
追踪文件维护
流程衔接

Key facts

Author: zenstory-ai
License: MIT (README badge)
Installs on skills.sh: 16,732 (Nov 2, 2026)
GitHub stars: 8,223 (Nov 2, 2026), whole repo
Files: 22 files including SKILL.md
Last commit: Nov 1, 2026 (repo)
Skill version: 1.1.1 in the frontmatter
Modes: full (4 agents), lean (2 agents), solo (no agents)
Platform rubrics: Fanqie, Qidian, Zhihu, plus a generic one
Bundled scripts: 6 scripts: 4 JavaScript and 2 Python

What it includes

Four reviewer agents in parallel
Story architect, character designer, narrative writer and consistency checker each review the draft, then results are merged and de-duplicated.
Runs code · /story-review (full)
Two reviewers: architect and consistency checker
Quicker, narrower review.
Files · /story-review lean
One session, no helper agents
Used on request or when agents are missing, broken or the tool is unavailable.
Read · /story-review solo
Punctuation, AI-pattern and degeneration checks
Runs normalize-punctuation.js --check, check-ai-patterns.js --check and check-degeneration.js --check on the chapter files. SKILL.md says these three report only and do not edit the chapters; their findings are merged into the report.
Runs code · Pre-check scripts
Must change, should change, may skip
S1 and S2 become must change, S3 should change, S4 may skip; each finding quotes the text.
Severity report
Revises 追踪/ files after a full or lean review
On long novels with a tracking state file, reads review-tracking.md to update tracking. Solo mode does not change tracking.
Write · Tracking update

From the official docs

Permissions

Low risk

What it can do

Read your data
Create and change things
Delete things
Read-only modeHow to turn it on

Watch out for

It runs bundled Node and Python scripts
Six scripts ship in the skill folder (4 JavaScript, 2 Python). SKILL.md runs three of them with --check before a review; the Python tracking_commit.py script is part of the tracking maintenance route and its source deletes files in some code paths. Review them if your drafts are sensitive. github.com
Full and lean reviews can update tracking files
On long novels with a tracking state file the review revises the 追踪 folder. Commit your book first. github.com
Batch reviews write a state file
For reviews split into batches, SKILL.md says every mode may write .story-review/state.md in your project. Solo mode writes nothing else to the project. github.com

Technical details

Turning on read-only	Use solo mode; the skill says solo does not change tracking files.

Questions

What does story-review do?

It reviews a web novel draft for structure, characters, prose, consistency and platform fit, and writes a report for the author. Findings quote the original text and are sorted into must change, should change and may skip.

How do I install it?

Use the install box above. For full and lean modes the helper agents must be deployed, so run /story-setup in your writing folder and start a new session. Without them it runs in solo mode.

What is the difference between full, lean and solo?

Full starts four reviewer agents in parallel, lean starts two, and solo uses only the current session. Any missing or failing agent drops the run to solo.

Does it change my chapters?

SKILL.md says the review does not modify chapter, setting or outline files, and that its three pre-check scripts only report. Full and lean reviews of a long novel with a tracking state file may revise the 追踪 tracking files, and batch reviews (many chapters) write a .story-review/state.md file. If you want fixes applied, SKILL.md points you to /story-deslop. Separately, the punctuation script's source has a code path that writes files; the review does not call it that way.

Does it need an API key?

The SKILL.md names no key. It uses your coding agent's own model and runs bundled Node and Python scripts.

Is it safe?

It reads your drafts, runs bundled scripts and may write tracking files in long-novel projects. Keep your book under version control so any tracking change can be undone.

Which platforms does it judge by?

It has rubrics for Fanqie, Qidian and Zhihu, and a generic rubric as the default.

Information

updated Nov 2, 2026

Repository: zenstory-ai/oh-story-claudecode
Folder: skills/story-review

zenstory-ai/oh-story-claudecode skills.sh

Sources

Docs, issue trackers and discussions this page is built from.

Show all 3