Guide · Agent skill
How to use the story-short-scan agent skill (短篇网文扫榜)
Updated · written with AI from the project's own documentation and checked against our catalogue data
story-short-scan is an agent skill for writers and editors who work on Chinese short web fiction. It tells the AI agent how to gather popular short stories from platforms such as 点众 and 黑岩, group them by emotional effect and genre, and write a report on which directions are trending. This guide explains what the skill does, how to install it, how to trigger it, and what it needs, so you can run your first scan and match the findings against your own material.
短篇网文扫榜zenstory-ai/oh-story-claudecode · MIT Download the skill Quick start
- 1Download and unzip. Download the story-short-scan ZIP from this page and unzip it into ~/.claude/skills/ or the project's .claude/skills/ folder. In the Claude apps, upload it under Settings, then Capabilities, then Skills.
- 2Install prerequisites. Install Node.js to run the bundled scripts. For scripted collection, also set up the separate browser-cdp skill so Chrome can run with remote debugging.
- 3Prepare the platforms. For 点众, no login is needed. For 黑岩, log in to the 黑岩 backend account in the Chrome window. For 知乎盐言, 番茄 and 七猫, prepare screenshots, text or links of the rankings.
- 4Start a request. Ask in plain language, such as 短篇什么火, or use /短篇扫榜. Say which platform and whether you have a genre in mind.
- 5Let the agent collect and aggregate. The agent runs the scrapers into a new dated folder, then aggregates the results and checks the collection status for problems.
- 6Read the report. Open 短篇扫榜结论.md in the output folder. Check the sample size and confidence line before you act on the directions.
- 7Match to your plan. Give the agent your target platform, materials and planned word count. It writes a topic-match section that rates feasibility.
What the story-short-scan skill makes the agent do
The skill is a set of instructions and scripts that teaches the agent a fixed workflow for studying short fiction. It first decides which platform or direction to focus on, then collects ranking or listing data, aggregates it, and writes a report on emotional directions, topic candidates and how to verify them.
The skill treats short fiction as an emotion-driven market. It judges each story from its genre and blurb, asking what feeling the reader gets, what triggers it and where the emotional peak is. It does not judge story quality: the skill states that ranking and engagement numbers show attention, not quality.
- Collects data from 点众 and 黑岩 with bundled scripts when a Chrome environment is available
- Accepts manually supplied rankings (screenshots, text or links) for other platforms such as 知乎盐言, 番茄 and 七猫
- Groups stories into six emotional categories and compares sample counts
- Writes a dated report with a confidence level and a suggested re-scan date
- Matches findings against an editor's submission brief or your materials and word-count plan if you provide them
Install the skill from the ZIP
Download the story-short-scan ZIP from this directory page and unzip it. You can install it for yourself across all projects, or only for one project.
For a personal install, unzip the folder into ~/.claude/skills/. For a single project, put it in the project's .claude/skills/ folder. In the Claude apps you can instead upload the skill under Settings, then Capabilities, then Skills.
- Personal: ~/.claude/skills/story-short-scan/
- Project only: the project's .claude/skills/story-short-scan/
- Claude apps: Settings → Capabilities → Skills, then upload
~/.claude/skills/story-short-scan/Trigger the skill with a natural request
You can call the skill with its slash command or by describing what you want in plain Chinese. The skill is designed to start from these triggers: /story-short-scan, /短篇扫榜, 「短篇什么火」 and 「知乎故事排行」.
If you already have a direction, say so in the first message. The skill uses it directly and runs a deep scan on that direction. If you have no direction, ask for a full overview of the platform. If you want to compare platforms, ask for a cross-platform comparison.
- Full overview: 短篇什么火？ (what is hot in short fiction right now)
- Platform ranking: 知乎故事排行 (for a platform you supply data for)
- Direction scan: tell it the genre or emotional direction you care about
/短篇扫榜Choose a data source for each platform
The skill uses three modes, in priority order. Scripted collection is used first when Chrome is available. Author-supplied data covers platforms without a script. Built-in knowledge is a fallback when the agent cannot go online and you have no data.
Built-in knowledge output is labelled as historical trend data and is only a candidate hypothesis until it is checked against live rankings. It does not include current counts or trends.
- Scripted collection: 点众 and 黑岩 (needs Chrome and Node.js)
- Author-supplied: 知乎盐言, 番茄短篇, 七猫短篇 (send screenshots, text or links)
- Built-in knowledge: no live data; treat results as hypotheses to verify
Run the scripted collection and aggregation
For scripted collection, the agent starts Chrome with remote debugging through the separate browser-cdp skill, then runs the platform scrapers. The 点众 scraper needs no login. The 黑岩 scraper needs you to log in to the 黑岩 backend in that Chrome window first. Each run writes to a new dated folder under 扫榜/, and the agent then aggregates the results into a summary file.
The main session reads only the aggregated summary. If the summary's collection status shows problems, the agent re-collects those platforms or says in the report which platform was left out.
- Each scan creates a new folder named by date (for example 扫榜/YYYYMMDD/); a same-day re-scan adds -2
- Do not add new results to an old folder, because the aggregation would mix in old rankings
- If a platform fails, the agent skips it and reports that
- For author-supplied rankings, the agent saves them in the same output folder using the platform reference format
node scripts/dz-browse-scraper.js --channel male --outdir {输出目录}
node scripts/heiyan-booklist-scraper.js --channel all --pages 3 --detail --outdir {输出目录}
node scripts/aggregate-rank.js {输出目录} --out {输出目录}/扫榜聚合.md --sparse 10 --scale shortRead the report and pick a direction
The report is written for the author and saved as 短篇扫榜结论.md in the output folder. It covers the market overview, an emotion ranking, topic hotspots, key data points such as word-count range and blurb style, and a list of directions worth writing. Sections that lack data are left out, and a single-sample scan has no change section.
For 点众 and 黑岩, the data shows titles currently on the shelf, not a heat ranking. The report labels these counts as shelf counts and does not say which story is popular.
- Compare numbers only within the same metric (the skill notes that each platform defines favourites, likes, reading counts and ratings differently)
- Treat the confidence line as the limit on how far to trust the result
- Re-scan around the suggested date, because trend signals can fade within weeks
Match the findings to your story plan
If you have an editor's brief or your own materials, give them to the agent. It saves the brief as 征稿要求.md, treats the genre, word count and deadline as hard requirements, and then writes a topic-match section into the same report file.
The agent asks for any missing information in one question: which platform you plan to submit to (or whether you will self-publish), what material and writing strengths you have, and how many words you plan to write. If you do not answer, it gives directions based on the sample and rates feasibility as medium.
- Each suggested topic states the emotional experience it gives the reader before it discusses the topic
- The first three sentences are the highest retention risk, so they should set up conflict, an identity gap or an emotional hook
- Reversals are a common spreading mechanism in short fiction; if you skip one, the skill says you need strong resonance, topic or aftertaste to compensate
What people use it for
- A writer with no topic picks a direction from the emotion ranking and a few sample titles before drafting
- An editor compares current shelf listings on 点众 male and female channels by genre and word count
- A writer checks an editor's submission brief against recent platform listings to see whether the requested theme is common
- A content planner collects manually supplied rankings from several platforms and compares which emotional directions appear across them
Frequently asked questions
Is the story-short-scan skill free to use?
The source for this skill does not state pricing or licence terms beyond the MIT licence shown on this page. Check the upstream repository's licence file for the exact terms, and note that the Chrome, Node.js and any platform accounts you use are separate requirements.
What do I need to install and run story-short-scan?
You need Node.js to run the bundled scripts. For automatic collection from 点众 and 黑岩, you also need Chrome with remote debugging enabled through the separate browser-cdp skill. A 黑岩 backend account is needed for the 黑岩 scraper. Other platforms require you to supply the ranking data yourself.
Where do I put the skill folder after downloading?
Unzip it into ~/.claude/skills/ for a personal install, or into the project's .claude/skills/ folder for a single project. In the Claude apps, you can upload it under Settings, then Capabilities, then Skills.
Does the skill work for platforms other than 点众 and 黑岩?
Yes, but without a script. For 知乎盐言, 番茄短篇 and 七猫短篇, you send the agent screenshots, text or links of the rankings, and it organises them into a file and analyses them. Reports from this path have a lower confidence level if you give only story titles.
Why does the skill say its trend data may be wrong?
The skill says trend signals can fade within weeks and that built-in historical data is only a hypothesis until it is checked against live rankings. Use the report's confidence line and suggested re-scan date to decide how much to rely on it.
Can the skill tell me whether a story is good?
No. The skill states that ranking and engagement numbers measure attention on a platform, not story quality or how well a story pulls the reader along. Use it to choose a direction, then judge your draft yourself.
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-short-scan if you write or edit Chinese short web fiction and want a dated, evidence-labelled read on which emotional directions and genres are showing up on specific platforms before you draft. Pick a different tool if you need a verified real-time forecast, if you work outside the Chinese web-fiction market, or if you want an automatic judgement of story quality.
Download the skill