churn-risk-playbook
GitHub指导如何结合产品遥测数据与流失风险模拟，生成可解释的客户流失风险评估。要求按顺序调用工具，将概率、关键驱动因素及警报结合，避免黑盒输出，提供包含ARR、健康度等指标的风险摘要。
触发场景
安装
npx skills add schmitech/orbit --skill churn-risk-playbook -g -y
SKILL.md
Frontmatter
{
"name": "churn-risk-playbook",
"enabled": true,
"version": "1.0",
"priority": 0,
"mcp_tools": [
"business-sample__get_product_telemetry",
"business-sample__simulate_churn_risk_scenario"
],
"description": "How to combine product telemetry with churn-risk simulation for a customer."
}
Order of calls
Call get_product_telemetry for the customer before
simulate_churn_risk_scenario. Both tools use the server's current customer
data; the telemetry result supplies the adoption context needed to explain the
simulation rather than relying on model assumptions.
Presenting results
Never present a bare churn probability. Always pair it with the simulation's key drivers and the telemetry's utilization or adoption alerts, so the number is explainable rather than a black box. The simulation itself accounts for open P1/P2 support escalations; telemetry does not return ticket volume.
Render a concise risk summary with Current ARR, Health Score, Seat
Utilization, Churn Probability, ARR at Risk, and Risk Category. Follow it with
the simulation's keyDrivers and the telemetry's telemetryAlerts. Do not
infer support-ticket volume beyond the simulation's openEscalationsCount, or
invent a trend or risk driver that neither result returns.
版本历史
- c378b23 当前 2026-09-09 13:54
