marsloting
User7 self-contained product-thinking skills that make an AI agent reason like a sharp PM before it acts. MIT.
Categories
Indexed Skills (7)
decision-matrix
Apply Pugh matrix + Multi-Criteria Decision Analysis (MCDA) when choosing among multiple options across multiple weighted criteria. Use whenever 用户 says "几个方案选哪个" / "怎么决定" / "团队意见不一致" / "vendor 选型" / "技术选型" / "供应商对比" / "decision matrix" / "比较 X 和 Y 和 Z". Forces criteria definition with explicit weights, scores each option per criterion, identifies dominant solutions, surfaces hidden disagreements (when team scores diverge significantly). Stops "感觉 X 比较好" decision by structure. Distinct from RICE (RICE = single-dimension prioritization, this = multi-criteria comparison). Self-contained methodology — no external docs required.
journey-mapping
Map customer journey + service blueprint when looking at end-to-end user experience, finding drop-off points, designing onboarding, or auditing existing flows. Use whenever 用户 says "用户体验全流程" / "哪一步流失最多" / "onboarding 重设计" / "service blueprint" / "customer journey" / "用户路径" / "全链路体验" / "frontstage backstage". Forces stage-by-stage breakdown (Awareness/Consideration/Onboarding/Use/Retention/Advocacy), captures Actions + Touchpoints + Emotions + Pain Points + Opportunities at each stage, separates frontstage (user-facing) from backstage (internal processes / systems / handoffs). Stops "feature thinking" by forcing whole-flow view. Self-contained methodology — no external docs required.
jtbd-framework
Apply Jobs-to-be-Done framework + Persona + Empathy map when uncovering what users actually want vs what they say they want. Use whenever 用户 says "用户到底想要什么" / "我们解决的是什么问题" / "persona 怎么定义" / "用户研究" / "JTBD" / "user need" / "为什么用户用我们" / "为什么不用". Forces functional + emotional + social job dimensions, distinguishes user-stated wants from underlying jobs, generates job statement in "When __ I want to __ so I can __" format, builds empathy map (Says/Thinks/Does/Feels) to triangulate. Stops feature-thinking, forces job-thinking. Self-contained methodology — no external docs required.
pre-mortem
Run pre-mortem (Klein) and FMEA-style failure-mode analysis BEFORE shipping a plan, feature, launch, campaign, or strategic decision. Use whenever 用户 says "这个方案有什么风险" / "上线前最后过一遍" / "万一失败了什么原因" / "risk register" / "what could go wrong" / "事先复盘" / "失败推演". Forces "imagine it failed in 6 months — write the autopsy" perspective shift, distinguishes pre-mortem (causes) from FMEA (severity × probability × detectability), produces concrete risk register with mitigation owner. Stops optimism bias, surfaces blind spots before they ship. Self-contained methodology — no external docs required.
rice-prioritization
Apply RICE / ICE / WSJF / MoSCoW prioritization frameworks when ranking multiple feature ideas, backlog items, experiments, or initiatives. Use whenever 用户 asks "哪个先做" / "这个值不值得" / "怎么排序" / "优先级排一下" / "rank these" / "should we do X first". Forces explicit scoring on Reach × Impact × Confidence ÷ Effort (RICE) or job size / time-criticality / risk-reduction (WSJF). Stops gut-feel prioritization, forces dimension-by-dimension reasoning, surfaces hidden Confidence assumptions, separates "feels important" from "scores high". Self-contained methodology — no external docs required.
root-cause
Apply 5 Whys + Fishbone (Ishikawa) + Pareto when something already broke and you need to find the root cause, not the symptom. Use whenever 用户 says "为什么会这样" / "找根因" / "这个 bug 怎么发生的" / "用户为什么流失" / "数据为什么跌" / "上次也踩过这个坑" / "incident postmortem" / "事故复盘". Forces 5-level "why" probing past the first surface answer, distinguishes proximate cause from root cause, uses Fishbone's 6M categories (man/machine/method/material/measurement/environment) for systemic issues, applies Pareto 80/20 to prioritize fixes. Stops "fix the symptom" pattern. Self-contained methodology — no external docs required.
story-splitting
Apply INVEST + SPIDR + 7 splitting strategies when a feature, requirement, or epic is too large to ship in one PR / one sprint / one week. Use whenever 用户 says "这个需求太大" / "拆一下" / "怎么分阶段做" / "acceptance criteria 怎么写" / "edge case 漏没漏" / "spec 写不下去" / "split this epic" / "break down this story". Forces INVEST checking (Independent / Negotiable / Valuable / Estimable / Small / Testable), produces vertical slices not horizontal layers, drafts acceptance criteria in Given-When-Then, enumerates edge cases by 4 axes (input / state / boundary / failure). Self-contained methodology — no external docs required.
Bio shown is the top-scored skill's repo description as a fallback — real GitHub bios land in a future update.