Skip to main content
ClaudeWave
Skill82 repo starsupdated 2d ago

wjs-x-increasing-follower

Use when 王建硕 wants to systematically grow his X (Twitter) followers by running numbered, A/B-testable growth experiments and tracking which ones actually work. Every action gets a number, a hypothesis, a target metric, a before-state (for rollback), and a verdict. The North-Star metric is the new-follower ÷ profile-visit ratio (conversion) — every profile change is judged against it. Daily it ingests the X Analytics CSV export, scores each running experiment, and recommends keep / rollback. Triggers — "涨粉", "增加 X 粉丝", "X 涨粉实验", "follower growth", "A/B test my profile", "今天的涨粉检查", "/wjs-x-increasing-follower".

Install in Claude Code
Copy
git clone --depth 1 https://github.com/jianshuo/claude-skills /tmp/wjs-x-increasing-follower && cp -r /tmp/wjs-x-increasing-follower/wjs-x-increasing-follower ~/.claude/skills/wjs-x-increasing-follower
Then start a new Claude Code session; the skill loads automatically.

SKILL.md

# wjs-x-increasing-follower

把「涨粉」当工程做:**每个改动是一个带编号的实验**,有假设、有目标指标、有 before(可回滚)、有判决。不靠感觉,靠数。

## Core Principle

**单账号没法做平行 A/B —— 只能做时间轴上的前后对比。** 所以唯一可信的北极星指标是 **转化率 = 新增关注 ÷ 主页访问(ratio)**:它对爆款流量免疫。一条推爆了带来一堆访问,ratio 不一定动;但 bio 改好了,每个来访的人更愿意关注,ratio 一定动。

**所以指标分层(每个 action 必须声明自己被哪个指标考核):**

| Action 类型 | 拨动的杠杆 | 用什么考核 |
|---|---|---|
| **profile**(bio / 名字 / 头像 / banner / 置顶 / URL / 地点) | 转化 | **ratio**(北极星,抗爆款) |
| **posting**(格式 / 钩子 / 频率 / thread vs 单条) | 触达 | profile visits + impressions(ratio 当护栏,别把转化拖垮) |
| **engagement**(回复 / 关注别人 / 互动) | 触达 | new follows + visits |
| **timing**(发布时间) | 触达 | profile visits |

**诚实护栏(写死在 evaluate.py 里):** 用中位数不用均值(一天爆款骗不了判决);够 7 天 / 够数据才下判决;同指标实验窗口重叠 → 打 `confounded` 标记;只给「方向性」结论,不号称因果。

**回滚是一等公民。** 每个 action 存了精确的 before 值,#N 永远能还原。判 ❌ rollback **先问王建硕,绝不静默改他的 bio**。

## 数据从哪来(关键约束)

ratio 这个数 **X API / xurl 拿不到** —— 只活在 Analytics 看板里。所以靠 **CSV 导出**:
打开 `x.com/i/account_analytics` → Overview → 右上角下载图标(7D/2W/4W/3M/1Y 想要哪段先选好)→ 导出 CSV → 丢进 `inbox/`(或直接给路径)。

## When This Skill Fires

- 王建硕说「涨粉」「搞个 X 涨粉实验」「A/B 测一下我的 profile」「今天的涨粉检查」
- 跑 `/wjs-x-increasing-follower`
- 设了 `/schedule daily /wjs-x-increasing-follower`(见末尾「每日检查」)

## When NOT to use

- 只是要发一条推 → `/wjs-tweeting-from-articles` 或直接 `xurl post`
- 推广 skill → `/wjs-promoting-skills` / `/publish-skill`
- 把文章分发到各平台 → `/wjs-syndicating-articles`

---

## Workflow

所有脚本在 `scripts/`,状态在 `state/`。先 `cd` 到 skill 目录。

### Step 1 — 吃数据(CSV → daily.jsonl)

```bash
python3 scripts/ingest-csv.py /path/to/export.csv
```

脚本模糊匹配列名(Profile visits / New follows / Impressions / Followers),按日期 upsert,自动算 ratio。**它会打印它认的列映射 —— 看一眼对不对**,不对就 `--visits-col "X" --follows-col "Y"`。重跑更长的导出会覆盖重叠的天,安全。

### Step 2 — 出 to-do(提实验)

看 `state/daily.jsonl` 现状,给王建硕一份**带编号的实验菜单**。每条必须有:`category` / `title` / `hypothesis` / `metric`。原则:

- **一次只让少数 profile 实验在跑**(同指标重叠会互相污染读数)。
- 假设要可证伪:「concrete 一行证明比模糊 tagline 转化高」✅;「让 bio 更好」❌。
- profile 类先于 posting 类 —— 转化是地基,先把来访的人接住,再去放大流量。

入账(自动分配编号,别手改 jsonl):

```bash
python3 scripts/ledger.py add --category profile --metric ratio \
  --title "Bio: 开头一行放硬证明" \
  --hypothesis "访客看到具体成绩比看到口号更愿意关注"
```

实验菜单参考(按杠杆分):
- **profile/ratio**:bio 重写(钩子前置 / 硬证明 / CTA「关注看 X」)、显示名加身份锚、头像换正脸高清、banner 放一句话价值主张、置顶换最强 thread、URL 指向落地页。
- **posting/visits**:发布格式(thread vs 单条)、首句钩子、带图 vs 纯文、话题聚焦。
- **engagement/follows**:固定回复某圈层、主动关注目标人群、参与热门话题。
- **timing/visits**:早 9 点 vs 晚 9 点、工作日 vs 周末。

`state/SCOREBOARD.md` 的「To-do」区就是这份清单的落地版(见 Step 6)。

### Step 3 — 上线一个实验(含抓 before)

**先抓 before-state**(profile 字段):

```bash
xurl "/2/users/me?user.fields=name,description,url,location,profile_image_url,pinned_tweet_id"
```

记下当前值。**然后真正去改:**

- **bio / 名字 / URL / 地点** —— 可走 API(OAuth1 v1.1):
  ```bash
  xurl --auth oauth1 -X POST "/1.1/account/update_profile.json?description=<urlencoded>"
  ```
  (`name` / `url` / `location` 同理。失败/没 enroll OAuth1 → 退回 App 里手改,照样记账。)
- **头像 / banner / 置顶推** —— 没有可靠公开 API,**在 App / 网页里手改**,skill 只负责记 before/after + 跟踪。

**记账(盖上 applied 日期 + before 值):**

```bash
python3 scripts/ledger.py apply 1 \
  --before-field description --before-value "<原 bio 原文>" \
  --after-value "<新 bio>"
```

之后这个实验进入测量期:baseline = applied 前 `window_days`(默认7) 天的指标中位数;判决要等 applied 后攒够数据。

### Step 4 — 每天喂数 + 重算(见也用于「每日检查」)

```bash
scripts/daily-check.sh
```

它会:ingest `inbox/` 里的新 CSV(用完归档到 `inbox/done/`)→ `evaluate.py --write-verdict` 把方向性判决写回账本 → 重生成 `SCOREBOARD.md`。**注意:它只写判决,不动状态** —— keep/rollback 是人的决定。

### Step 5 — 判决与回滚

```bash
python3 scripts/evaluate.py          # 看每个在跑实验:baseline vs post,Δ%,verdict
```

判决(相对 baseline 中位数):`Δ ≥ +10%` → **keep**;`Δ ≤ -10%` → **rollback**;之间 → **flat**。注意 `⚠ confounded` / `⚠ thin baseline` 标记,有就别太当真,延长窗口或停掉重叠实验再测。

- **keep**:`python3 scripts/ledger.py keep 1 --note "+58% 转化"`
- **rollback(先问王建硕!)**:确认后,用 `apply` 时存的 before 值还原(API 或手改),再
  `python3 scripts/ledger.py rollback 1 --note "转化掉了,还原 bio"`(命令会打印要还原的 before 值)
- **flat**:保留观察,或归为「不影响」撤掉。

### Step 6 — 看板

```bash
python3 scripts/scoreboard.py        # 写并打印 state/SCOREBOARD.md
```

四块:📈 现状(最新 ratio / 7 日中位 ratio / 粉丝数)、🧪 在跑实验(baseline/post/Δ/判决/告警)、✅ To-do 待办(proposed 清单)、📚 已结案(kept / rolled_back)。给王建硕看就发这个文件。

---

## 数据模型(state/)

- `daily.jsonl` —— 一天一行:`{date, profile_visits, new_follows, ratio, impressions, followers_total}`。
- `actions.jsonl` —— 一实验一行:`{id, category, title, hypothesis, metric, window_days, before:{field,value}, after:{value}, applied, status, evaluated, verdict, notes}`。状态流:`proposed → active → (kept | rolled_back)`。
- `SCOREBOARD.md` —— 生成物,人看的。

## 默认参数(要改就传 flag)

- 评估窗口 `--window-days` 默认 **7 天**
- keep/rollback 阈值 `--threshold` 默认 **±0.10(相对 ±10%)**
- 下判决前最少 post 数据 `--min-post-days` 默认 **3 天**
- 回滚 **永远先问**,不静默

## 每日检查(可选调度)

`/schedule daily /wjs-x-increasing-follower` —— 每天调起本 skill:跑 `daily-check.sh`,把 `SCOREBOARD.md` 现状 + 任何判 ❌ rollback 的实验 **发给王建硕并征求是否回滚**。CSV 还是得他手动导出丢进 `inbox/`(ratio 数据 API 拿不到,这步绕不开)。
skill-quality-reviewerSubagent

Repo-wide drift detector for the wjs-* Claude Code skills in this marketplace. Sweeps every SKILL.md, scores it against the repo's own conventions (V-ing naming, trigger-phrase density, companion files, description shape), and returns a grouped punch list ordered by severity. Read-only — never edits files. Use before pushing a batch of skill changes, or whenever you wonder "are these skills still internally consistent?

wangjianshuo-perspectiveSkill

|

wjs-auditing-projectSkill

Use when the user asks to audit what's wrong with a project, "make it right", "看看项目出了什么问题", "为什么用户的需求还没上线", "为什么没提交App Store", "为什么没新build", or wants a holistic state-of-the-project check covering unmerged branches, stalled PRs, failed GitHub Actions, stale builds, plan drift (TODOS.md / ROADMAP), unreleased commits, and log errors. Runs read-only investigation, presents a grouped checklist, fixes only after explicit user confirmation. Aware of the Cathier iOS app workflow (Xcode + fastlane + auto-merge @claude PRs from in-app feedback).

wjs-burning-subtitlesSkill

Use when the user has a video + an SRT and wants the subtitles either burned into the pixels (libass, always-visible) or soft-muxed as a togglable track. Also handles the final composite step for the localization pipeline — burn subs, mix a dub track, and keep the original audio as a low-volume bed, all in ONE ffmpeg encode (no cascade). Verifies libass availability and auto-downloads a static evermeet ffmpeg build when Homebrew's stripped binary lacks it. Triggers — "烧字幕", "硬字幕", "burn subtitles", "burn-in subs", "embed subtitle", "soft mux SRT", "把字幕烧进视频", "做最终合成".

wjs-cleaning-spamSkill

Use when the user complains about spam on his X/Twitter posts — 同城面付 / 寻固炮 / 线下上门 / 免费破处 这类引流号在他推文下刷的 emoji 垃圾回复 — and wants them removed. Covers the last 7 days (X recent-search window). Triggers — "把这些spam删掉", "清理X垃圾回复", "推文下面好多引流号", "clean spam replies", "/wjs-cleaning-spam".

wjs-converting-text-to-videoSkill

Use when the user wants a 王建硕-style WeChat article (article.md) turned into a narrated short MP4 video — TTS voiceover via 火山引擎 Volcano TTS, HyperFrames CSS/GSAP animation per scene, subtle SFX, abstract watercolor background, full pipeline rendering to 1080×1920 portrait MP4 (30-90s). Triggers — "把这篇文章做成视频", "做一个解说视频", "讲解视频", "/wjs-converting-text-to-video".

wjs-converting-wp-to-hugoSkill

Use when migrating a WordPress site to a Hugo static site on GitHub Pages from a WXR export (.xml) plus the wp-content/uploads folder — preserving /archives/<id>/ URLs, localizing images, and deploying via GitHub Actions. Triggers — "把 WordPress 迁成 Hugo", "wordpress 转静态站", "migrate WordPress to Hugo", "WXR to Hugo", "publish WordPress to GitHub Pages", "/wjs-converting-wp-to-hugo".

wjs-dubbing-videoSkill

Use when the user has a video + a target-language SRT and wants the video to actually speak that language — generates a time-aligned TTS voice dub. Routes by voice ID — Volcano (豆包) TTS for Chinese, edge-tts neural for any language. Defaults to one voice (single-speaker); opt-in multi-speaker via visual diarization. Outputs `*_<lang>_dub.mp4` with the dub audio in place of the original. Final mixing (audio bed + burn-in) is handed off to `/wjs-burning-subtitles`. Triggers — "配音", "中文配音", "Chinese dub", "voice over this", "dub the video", "TTS this SRT", "different voice for each speaker".