Agent Skills
› mtarcure/claude-vibe-squad
› virality-analysis
virality-analysis
GitHub用于分析内容传播原因或设计平台分享假设测试。定义受众、指标、替代方案和停止规则，生成基于证据的实验记录，严禁臆造数据或越权访问付费/私密分析数据。
Trigger Scenarios
分析特定内容为何扩散
设计平台特定的分享假设测试
Install
npx skills add mtarcure/claude-vibe-squad --skill virality-analysis -g -y
SKILL.md
Frontmatter
{
"name": "virality-analysis",
"audience": "specialist",
"description": "Use when analyzing why specific content spread or designing a platform-specific test of a sharing hypothesis—define cohorts, metrics, alternatives, confidence, and a stopping rule from observed data. Do not claim causality or access paid\/credentialed analytics without the required gates."
}
Virality Analysis
Analyze a platform-specific sharing hypothesis — hook, retention, shareability, and loop — and turn it into an evidence-bounded experiment rather than a causal story.
Required experiment record
For each hypothesis, record the platform, source, observation window, cohort definition, metric names and formulas, alternative hypotheses, current confidence, intervention/test, stopping rule, and result status. Also state whether the test needs paid distribution, credentials, or private audience/analytics data; those requirements stay behind their separate operator, budget, and privacy gates.
Steps
- Pin the platform, evidence source, observation window, cohort, and metric definitions before interpreting signals.
- Identify a candidate share driver: the emotion or utility that might make a person pass it on.
- Analyze hook and retention structure against the platform mechanics observed in the same window.
- Map the sharing/growth loop and state at least one alternative explanation for every proposed mechanism.
- Assign confidence from observed signals, never fabricated views, retention, or engagement rates.
- Define a testable intervention and stopping rule; mark gated or unavailable analytics as unmeasured, not inferred.
- Update the experiment record with the outcome and retain alternatives that the result did not distinguish.
Acceptance
- Every recommendation is tied to one platform-specific experiment record and mechanism.
- Source/window/cohort and metric formulas are explicit; alternatives and confidence are recorded.
- A test and stopping rule exist before results are interpreted.
- No causal claim exceeds the experiment, no metric is fabricated, and paid/credential/private-data needs remain gated.
Version History
- d5262e2 Current 2026-09-11 11:46
