技能备份 - 2026-04-15 (40个技能)

2026-04-15 18:53:15 +08:00
parent f62f14814f
commit c65fce24e4
791 changed files with 190773 additions and 0 deletions
@@ -0,0 +1,77 @@
+---
+name: Product Manager
+description: Build products users love with discovery, prioritization, roadmapping, and cross-functional leadership.
+metadata: {"clawdbot":{"emoji":"🎯","os":["linux","darwin","win32"]}}
+---
+
+# Product Management Rules
+
+## Discovery
+- Talk to users weekly — not just at project kickoff
+- Watch behavior, don't just collect opinions — users say one thing, do another
+- Problem validation before solution validation — are we solving the right thing?
+- Jobs to be done: what's the user trying to accomplish?
+- Competitors show what's possible, not what to copy
+
+## Prioritization
+- Impact vs effort is a starting point, not the answer
+- Say no more than yes — focus is a feature
+- Urgent vs important: stakeholder pressure isn't priority
+- Stack rank ruthlessly — "everything is P1" means nothing is
+- Revisit priorities when context changes — quarterly at minimum
+
+## Roadmapping
+- Outcomes over outputs — what will change, not what we'll build
+- Time horizons: now (committed), next (planned), later (possible)
+- Communicate uncertainty honestly — roadmaps aren't promises
+- Dependencies surfaced early — blocked work wastes everyone's time
+- Update when reality changes — stale roadmaps destroy trust
+
+## Requirements
+- User stories: who, what, why — not how
+- Acceptance criteria define done — ambiguity creates rework
+- Edge cases addressed upfront — not discovered in QA
+- Scope creep is the enemy — good enough now beats perfect later
+- Technical constraints are real — work with engineering, not around them
+
+## Working with Engineering
+- Context over directives — explain why, not just what
+- Tradeoffs are collaborative decisions
+- Spec before sprint, not during — no designing on the fly
+- Protect focus time — meetings kill flow
+- Trust their estimates, push back on scope not time
+
+## Working with Design
+- Research together, don't hand off briefs
+- Critique the work, not the designer
+- Design reviews with users, not just stakeholders
+- Mobile and edge cases early — not afterthoughts
+- Design system enables speed — support it
+
+## Stakeholder Management
+- Regular updates prevent surprise requests
+- Data calms opinion battles
+- Explain trade-offs, don't just defend decisions
+- Feedback channels prevent end-runs — make input easy
+- Executive sponsors for big initiatives
+
+## Metrics
+- One north star metric, 2-3 supporting
+- Leading indicators for early signal — don't wait for lagging
+- Dashboards should prompt questions, not just display numbers
+- Vanity metrics feel good, don't drive decisions
+- A/B test when data beats intuition
+
+## Launch
+- Soft launch catches problems before scale
+- Success criteria defined before launch — not after
+- Rollback plan before rollout
+- Cross-functional checklist: docs, support, marketing
+- Post-launch review: what worked, what didn't
+
+## Common Mistakes
+- Feature factory: shipping without learning
+- Overspeccing: killing engineering autonomy
+- Consensus seeking: decisions by committee
+- Ignoring qualitative: data alone misses why
+- Roadmap as backlog: detail everything, commit nothing
@@ -0,0 +1,6 @@
+{
+  "ownerId": "kn73vp5rarc3b14rc7wjcw8f8580t5d1",
+  "slug": "product-manager",
+  "version": "1.0.0",
+  "publishedAt": 1770675169581
+}