Agent Skills
› NeverSight/learn-skills.dev
› product-improvement-planner
product-improvement-planner
GitHub用于分析产品现状、映射用户旅程并生成改进创意。通过评估影响、风险和可行性进行优先级排序，制定详细的实施计划、验收标准及度量指标，确保方案安全合规且具备可执行性。
Trigger Scenarios
请求产品策略或路线图
发起产品改进冲刺规划
新功能头脑风暴
Install
npx skills add NeverSight/learn-skills.dev --skill product-improvement-planner -g -y
SKILL.md
Frontmatter
{
"name": "product-improvement-planner",
"description": "Use to analyze the current product, map the user journey, generate improvement ideas, prioritize by impact, effort, and risk, and plan implementation workflows. Trigger when the user requests product strategy, a roadmap, a product improvement sprint, or brainstorming for new features."
}
- Audit the current product surface. Review onboarding, practice, dashboard, AI tutor, review queue, settings, payments if present, analytics, support channels, and recent user feedback.
- Map the user journey from first visit through repeated use and target outcome. Identify friction points, unclear value, drop-off moments, manual workarounds, and trust gaps.
- Gather evidence before ideating where practical. Use product metrics, support notes, user interviews, competitor research, domain constraints, and accessibility or performance findings.
- Generate improvement ideas across onboarding, adaptive study plans, analytics, feedback loops, content quality, review scheduling, AI assistance, reliability, and retention.
- Score ideas by user impact, implementation effort, technical risk, business value, and reversibility. Make assumptions explicit when evidence is incomplete.
- Select the top few improvements and choose one primary plan unless the user asks for a broader roadmap. Explain why lower-ranked ideas are deferred.
- Create an execution plan for the top improvement covering design, data model changes, implementation steps, tests, rollout, analytics, support, and rollback.
- Define acceptance criteria and success metrics. Include both product metrics and engineering verification such as tests, accessibility, performance, and build health.
- Preserve safety and trust. Do not propose dark patterns, misleading readiness claims, unlicensed content use, privacy-invasive tracking, or medical advice.
- Finish with the prioritized list, selected plan, acceptance criteria, measurement plan, risks, and backlog items for future sprints.
Version History
- e0220ca Current 2026-07-05 22:11
