Guide · Agent skill
How to use the story-long-scan agent skill for web novel trends
Updated · written with AI from the project's own documentation and checked against our catalogue data
story-long-scan is an agent skill for Chinese long-form web fiction. It makes an AI agent pull ranking and new-book data from platforms such as Qidian, Fanqie, Jjwxc, Qimao and Ciweimao, then report which genres, character settings and title patterns keep recurring. This guide covers installing the skill, the requirements, how to trigger it with plain requests, and the limits you should know before relying on its output.
Long-form web novel market scanzenstory-ai/oh-story-claudecode · MIT Download the skill Quick start
- 1Download the ZIP. Download the story-long-scan ZIP from this page.
- 2Install the skill. Unzip it into ~/.claude/skills/ for personal use or .claude/skills/ for one project, or upload it under Settings → Capabilities → Skills in the Claude apps.
- 3Check Node.js and Chrome. Confirm Node.js is installed. For Fanqie, Jjwxc or Ciweimao, start Chrome with the browser-cdp skill first.
- 4Start a scan. Type /story-long-scan or a request such as 「起点排行」 and name the platform and channel.
- 5Answer the setup questions. The agent asks which platform, channel and focus you want, plus any genre you already have in mind.
- 6Read the report. Review the genre table, new signals and suggested directions, and check the collection notes for any missing rankings.
- 7Check the decision file. Open the 选题决策.md file in the dated folder to see the saved conclusion and recommended topics.
What an agent skill is and what this one makes the agent do
An agent skill is a folder of instructions and scripts that an AI agent loads when a task matches it. This skill is for market research before writing a long serial: it tells the agent how to collect rankings, group the results and write a report.
The skill's core rule is that one ranking position is only a clue. Trends are supported only when the same pattern repeats across several recent charts and many titles. The agent also weighs a topic against your material, genre limits, planned length and target platform, then suggests two or three topics to test.
- Collects ranking and new-book data from the platforms it supports
- Groups results by genre, tags, title words and blurb selling points
- Reports recurring patterns, rising or stable genres, and typical word counts
- Saves the report and topic decision to a file in your working folder
Install the story-long-scan skill
Download the ZIP for this skill from this page. You can install it in one of two ways. For a personal setup, unzip it into ~/.claude/skills/. For a single project, unzip it into the project's .claude/skills/ folder.
In the Claude apps you can instead upload the ZIP under Settings → Capabilities → Skills.
- Personal: unzip into ~/.claude/skills/
- Project only: unzip into .claude/skills/ inside the project folder
- Claude apps: Settings → Capabilities → Skills, then upload the ZIP
Check the requirements before you run a scan
The bundled scripts run on Node.js, so Node.js must be installed. The skill also needs network access to the platform websites.
Qidian rankings work without Chrome. Fanqie, Jjwxc, Ciweimao and other scraped platforms need Chrome running with the browser-cdp skill, using CDP on port 9222. Start it with /browser-cdp before scanning those platforms.
- Node.js, to run the scripts
- Network access to the platform websites
- Chrome plus the browser-cdp skill (CDP on port 9222) for Fanqie, Jjwxc, Ciweimao
- Qidian: no Chrome needed
Trigger the skill with natural requests
You can start a scan with the slash command or with a plain request. The skill recognises these triggers:
- /story-long-scan
- /长篇扫榜
- 「长篇什么火」(what long-form is trending)
- 「起点排行」(Qidian rankings)
- 「扫七猫新书」(scan Qimao new books)
Run a scan and read the output
The agent first asks which platform, which channel (male or female, for Fanqie and Qimao), and whether you want new books or established rankings. It then creates a dated folder named 扫榜/{YYYYMMDD}/ in your working directory and collects data into it. Once collection finishes, it aggregates the files into one summary with this command:
- If a ranking fails to collect, the agent checks the platform's notes and re-collects, or says in the report that the conclusion excludes it
- The report includes a genre ranking table, new genre signals, classic genre changes, new settings and plot patterns, word-count ranges and a shortlist of directions
node scripts/aggregate-rank.js {outdir} --out {outdir}/扫榜聚合.mdSafety notes: scripts, network and scraping
This skill runs scripts and uses the network. Review the scripts folder before you install it if that matters to you. The scrapers read web pages directly, so they depend on each site's current layout and anti-scraping measures. A layout change or a verification page can make a scraper return nothing or partial data.
The skill does not have access to private author dashboards. Completion rate, follow-up reading and subscription conversion cannot be measured by the scripts, so the agent will not draw conclusions about them unless you supply that data yourself. If live data is unavailable, the skill can fall back to built-in historical genre candidates, which may be out of date and must be checked before use.
- Scripts run on your machine; review them first if you need to
- Scrapers can break when page layouts change
- Private author metrics are not accessible to the scripts
- Offline fallback uses historical knowledge and may be outdated
Tips for better scans
Give the agent your target platform, your existing material, the genres you can write, and your planned length. These inputs decide which suggested topics are feasible. For a deeper look at one genre, you can ask the agent to sample raw entries from that genre, one genre at a time. The sample command takes a genre, tag or title word and a count.
- Use Qidian first if you want a scan without Chrome
- For new-book trends, include the new-book ranking rather than only the main charts
- Compare with an earlier dated folder only when it used the same platform, ranking and filters
- Treat small sample counts with caution; some platforms only show a few titles per ranking
node scripts/aggregate-rank.js {outdir} --sample {genre/tag/title word} --n 5What people use it for
- Choosing a genre for a new long serial on Qidian or Fanqie before drafting an outline
- Comparing which story types are rising on Qimao and Jjwxc in the same week
- Finding recurring character settings and title patterns to test in a first chapter
- Tracking genre trends over several dated scans for a market research report
Frequently asked questions
How do I install the story-long-scan skill?
Download the ZIP from this page and unzip it into ~/.claude/skills/ for personal use, or into .claude/skills/ inside a project. In the Claude apps, you can upload the ZIP under Settings → Capabilities → Skills instead.
What do I need to run story-long-scan?
You need Node.js to run the bundled scripts and network access to the platform websites. Fanqie, Jjwxc, Ciweimao and other scraped platforms also need Chrome running with the browser-cdp skill on port 9222. Qidian rankings work without Chrome.
Is story-long-scan free?
The skill is published under the MIT licence, as stated on this page. Check the licence file in the source repository for the full terms.
Why does a scan return empty or incomplete results?
Scrapers depend on each site's page layout and anti-scraping measures, so a layout change or a login or verification page can stop collection. The agent flags affected rankings, and you can re-run the scan after opening the site in Chrome to confirm it loads normally.
Can the skill show completion rate or subscription conversion?
No. Those numbers come from the author's backend, which the scripts cannot access. Supply them yourself if you have them, and the agent will use only that data.
How does story-long-scan compare with browsing rankings by hand?
The skill collects many titles across several rankings, groups them by genre and tag, and checks which patterns repeat. Manual browsing shows single chart positions, which the skill treats only as clues. It is built for Chinese web fiction platforms, so it does not cover other markets.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Use story-long-scan if you write or research Chinese web novels and want repeated genre and title patterns from several platforms without browsing each chart by hand. Pick something else if you work outside Chinese web fiction, need reliable live data without a browser session, or need private author metrics.
Download the skill