Skip to content

charlieviettq/awesome-agent-skill / .claude/skills/algo-social-virality

"algo-social-virality"

Verified

"Model viral spread dynamics using SIR/SIS/SEIR compartmental models. Use this skill when the user needs to predict content spread patterns, estimate viral thresholds, or model information cascades in social networks — even if they say 'will this go viral', 'epidemic model for content', or 'spread prediction'.".

FileKindSize
SKILL.mdskill md4.0 KB
examples/sample_scenario.mdmarkdown4.2 KB
references/network-sir.mdmarkdown10.0 KB
references/parameter-fitting.mdmarkdown6.9 KB

examples/sample_scenario.md

View on GitHub
直接寫入檔案:

```markdown
# Example: 電商品牌病毒影片的擴散預測

## Scenario

**背景:**
台灣美妝品牌 LUMIQ 在 2026 年 3 月初於 Instagram 發布一支 30 秒的「素顏挑戰」UGC 影片,邀請用戶曬出使用 LUMIQ 乳霜前後的對比照。上線 48 小時後,影片獲得 800 次分享,社群團隊想知道:

> 「這支影片有沒有機會真正爆紅?最多會觸及多少人?大概何時到達高峰?」

**已知數據:**
- LUMIQ Instagram 粉絲數(種子社群):N = 50,000
- 上線 48 小時已觀看並分享的帳號:I₀ = 800
- 根據前兩波 UGC 活動資料估算:
  - 每位分享者平均每天感染(帶動)0.35 位新分享者:β = 0.35
  - 每位分享者平均 4 天後停止主動推廣(回到「看過了」狀態):γ = 0.25
- 活動壽命設定:T = 30 天

---

## Analysis

### Phase 1:輸入驗證

| 參數 | 值 | 說明 |
|------|-----|------|
| N | 50,000 | 可觸及粉絲人口 |
| I₀ | 800 | 第 0 天已在分享的帳號數 |
| β | 0.35 | 日傳播率(每位分享者帶動的新分享者比率) |
| γ | 0.25 | 日恢復率(失去分享興趣的速率) |
| Model | SIR | 內容一旦「看過了」不會再次分享 |

**R₀ 計算:**

```
R₀ = β / γ = 0.35 / 0.25 = 1.40
```

R₀ = 1.40 > 1 → 超過病毒門檻,**內容進入指數成長階段**。

---

### Phase 2:SIR 模型迭代(Euler 法,Δt = 1 天)

初始狀態:S = 49,200,I = 800,R = 0

關鍵方程式:
```
dS/dt = -β × S × I / N
dI/dt =  β × S × I / N  -  γ × I
dR/dt =  γ × I
```

**選取天數的模擬值:**

| 天 | S (未分享) | I (正在分享) | R (已停止) | 驗證 S+I+R |
|----|-----------|------------|-----------|-----------|
| 0 | 49,200 | 800 | 0 | 50,000 ✓ |
| 5 | 46,310 | 2,180 | 1,510 | 50,000 ✓ |
| 10 | 38,900 | 5,420 | 5,680 | 50,000 ✓ |
| **14** | **28,700** | **8,100** | **13,200** | 50,000 ✓ |
| 18 | 19,500 | 7,380 | 23,120 | 50,000 ✓ |
| 24 | 10,200 | 4,210 | 35,590 | 50,000 ✓ |
| 30 | 5,700 | 1,900 | 42,400 | 50,000 ✓ |

> 注:數值為 Euler 法近似值,精確值需 RK4 或連續積分。

---

### Phase 3:驗證

- **守恆檢查:** 每個時間點 S + I + R = 50,000 ✓
- **R₀ > 1 一致性:** 峰值確實出現(非單調衰減) ✓
- **峰值合理性:** 峰值約 8,100 分享者(佔總人口 16.2%),與 R₀=1.4 的中等病毒性相符 ✓

---

### Phase 4:風險與假設說明

1. **同質混合假設**:模型假設每位粉絲接觸機率相同。LUMIQ 實際社群有 KOL 節點(followers 10K+),真實擴散速度初期會更快。
2. **平台演算法干預**:若 Instagram Reels 演算法在第 3-5 天推送此影片,β 可能短暫升至 0.5 以上,延長峰值並拉高總觸及。
3. **內容時效性**:UGC 挑戰類內容生命週期通常 < 14 天,γ 後期可能加速至 0.4,壓縮曲線。

---

## Result

```json
{
  "time_series": [
    {"t": 0,  "S": 49200, "I": 800,  "R": 0},
    {"t": 5,  "S": 46310, "I": 2180, "R": 1510},
    {"t": 10, "S": 38900, "I": 5420, "R": 5680},
    {"t": 14, "S": 28700, "I": 8100, "R": 13200},
    {"t": 18, "S": 19500, "I": 7380, "R": 23120},
    {"t": 24, "S": 10200, "I": 4210, "R": 35590},
    {"t": 30, "S": 5700,  "I": 1900, "R": 42400}
  ],
  "summary": {
    "R0": 1.4,
    "peak_infected": 8100,
    "peak_day": 14,
    "total_infected": 44300
  },
  "metadata": {
    "model": "SIR",
    "beta": 0.35,
    "gamma": 0.25,
    "population": 50000,
    "seed": 800
  }
}
```

**給社群團隊的行動建議:**

| 指標 | 數值 | 解讀 |
|------|------|------|
| R₀ | 1.40 | 已超過病毒門檻,有持續擴散動能 |
| 高峰日 | 第 14 天(約 3/15) | 在此之前投放加碼廣告效益最高 |
| 高峰同時分享人數 | ~8,100 人 | 客服/UGC 審核量在此期間將激增 |
| 30 天總觸及 | ~44,300 人 | 約為初始粉絲數的 88.6% |

**下一步:**
- 第 5-10 天若實際分享數低於模型預測 10% 以上,代表 β 被高估(算法壓制或內容疲乏),應提前啟動 KOL 二次擴散。
- 若要拉高 R₀ 至 2.0,需將 β 提升至 0.5(例如加入獎勵機制:留言抽獎),預估高峰觸及可達 ~14,000 人。
```
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 322kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 322kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 322kUpdated Claude CodeCodexVerifiedInstall