用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt;运行前请先检查来源。
npx skills add https://github.com/UitbreidenOS/UitKit --skill engineering-strategy命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地?可下载 SkillsMP 当前能够提供的文件。
正在显示 SKILL.md
Guidelines and instructions for Agent execution state rollback rules
Guidelines and instructions for Agent execution step counters limits
Guidelines and instructions for Agent execution timeout limits setups
基于 SOC 职业分类
| name | engineering-strategy |
| description | Engineering strategy document: tech vision, build vs buy decisions, team topology, 12-month roadmap |
/adr-writer for thoseWrite an engineering strategy document for [COMPANY].
Context:
- Company stage: [seed / series A / series B / growth / enterprise]
- Current engineering team size: [X engineers]
- Current architecture: [monolith / microservices / serverless / hybrid]
- Primary tech stack: [languages, frameworks, cloud provider]
- Current biggest technical pain: [e.g., deployment velocity, reliability, scaling, tech debt]
- Business context: [what the company is trying to achieve in the next 12 months]
- Top 3 product priorities from CEO/CPO: [list them]
Produce a strategy document covering:
## 1. Engineering Vision (12 months)
One paragraph: What does our engineering organisation look like in 12 months?
Specifically address:
- Deployment frequency (target)
- System reliability (uptime / error rate target)
- Team structure (how many teams, what model)
- Developer experience (how fast can a new engineer ship their first feature?)
## 2. Current State Assessment
Honest diagnosis — what's working, what's broken:
- Architecture: [current state and key limitations]
- Tech debt: [quantify if possible — % of dev time lost to it]
- Deployment velocity: [current deploys/day or week]
- Reliability: [current uptime, incident rate]
- Team structure: [current topology and where it breaks down]
## 3. Strategic Priorities (ranked)
Top 3-5 engineering bets for the next 12 months.
For each priority:
- What it is
- Why it matters (business impact, not technical elegance)
- What success looks like (measurable)
- Rough investment required (engineering weeks / headcount)
## 4. Build vs. Buy vs. Partner
For each major technical capability we need:
| Capability | Build | Buy | Partner | Recommendation | Rationale |
Use criteria:
- Core differentiator? → Build
- Commodity/solved problem? → Buy
- Need reach/network? → Partner
- Time to market critical? → Lean toward Buy
## 5. Team Topology
Current → Target structure over 12 months.
Team models to choose from:
- Stream-aligned teams (product-feature ownership)
- Platform/enabling teams (developer experience, infra)
- Complicated subsystem teams (ML, search, data pipeline)
Use Team Topologies vocabulary: stream-aligned, platform, enabling, complicated subsystem.
For each team: mission, size, tech ownership, interfaces to other teams.
## 6. Technology Bets
What are we committing to over the next 2-3 years?
- Core languages and frameworks (what we're standardising on)
- Cloud provider and key managed services
- What we're moving away from (sunset plan)
- What we're watching but not committing to yet
## 7. Engineering Health Metrics
How will we measure if the strategy is working?
| Metric | Current | 6-month target | 12-month target |
Include: DORA metrics (deployment frequency, lead time, MTTR, change failure rate), availability, developer NPS, tech debt ratio.
## 8. Risks and Mitigations
Top 3 risks to this strategy:
- Risk, likelihood, impact, mitigation
## 9. Investment Ask
What do we need to execute this strategy?
- Headcount: [X engineers to hire in next 12 months]
- Tooling budget: [$X for build-vs-buy decisions]
- Infrastructure: [expected infra cost change]
Help me decide whether to build or buy [CAPABILITY].
Capability description: [what we need it to do]
Our current approach: [how we handle it today, if at all]
Timeline pressure: [when we need it]
Engineering cost to build: [estimate in eng-weeks, or ask Claude to estimate]
Buy options I've identified: [vendor names, pricing if known]
Our team's expertise in this area: [strong / weak / none]
Evaluate against these criteria:
1. Core differentiator test
Is this capability part of our unique value proposition?
- YES → Strong signal to build (owning it = competitive moat)
- NO → Strong signal to buy (it's commodity infrastructure)
2. Complexity vs. expertise
- High complexity + low team expertise → Buy (build risk is high)
- High complexity + strong team expertise → Build (if differentiated)
- Low complexity + any expertise → Build (unless off-the-shelf is trivial)
3. Time to market
- Need in < 3 months → Buy almost always wins
- 3-12 months → depends on strategic importance
- 12+ months → build if differentiating
4. Total cost of ownership (3-year horizon)
Build: engineering cost + maintenance overhead + opportunity cost
Buy: licence fees + integration + lock-in premium
5. Vendor risk
- Startup vendor: lock-in risk, acquisition risk
- Established vendor: pricing power risk, slow roadmap risk
- Open source: maintenance burden, community risk
Output:
- Recommendation: Build / Buy / Hybrid / Delay
- 3 strongest reasons for the recommendation
- What would change your mind
- If Buy: vendor shortlist and next step
- If Build: rough architecture and team assignment
Design the team topology for our engineering organisation.
Current state:
- Total engineers: [X]
- Current teams: [list them and what they do]
- Biggest coordination problems: [where do handoffs break or slow things down?]
- Product areas: [list the major product domains]
- Platform/infra maturity: [strong / weak / non-existent]
Target state:
- Engineers in 12 months: [X (including hiring plan)]
- Primary business priority: [ship product features / scale infrastructure / reduce incidents]
Design the target topology using these team types:
1. Stream-aligned teams: Own a product domain end-to-end, fast flow, empowered
2. Platform team: Internal product — CI/CD, observability, developer tooling, infra
3. Enabling team: Temporary, coaches other teams through transitions (migration, new tech)
4. Complicated subsystem team: Deep expertise required — ML, search, payment processing
For each team in the target design:
- Team name and mission
- Team size (target and interim)
- What they own (services, features, infra)
- What they depend on (from platform team or external)
- How they interact with adjacent teams (API, shared service, consultation)
- Success metric for this team
Interaction modes between teams:
- Collaboration: works closely, frequent communication (temporary, for transitions)
- X-as-a-Service: consumer/provider relationship with defined interface
- Facilitating: one team helps another build capability (time-boxed)
Output: org chart + team charters + interaction model diagram (text-based)
Build a 12-month engineering roadmap.
Business priorities from leadership:
Q1: [what the company needs to ship / achieve]
Q2: [what the company needs to ship / achieve]
Q3: [what the company needs to ship / achieve]
Q4: [what the company needs to ship / achieve]
Engineering constraints:
- Current team capacity: [X engineers × 10 productive days/sprint]
- Planned hiring: [when and what roles]
- Known tech debt obligations: [what must be addressed]
- Planned migrations: [e.g., moving to microservices, upgrading infra]
Roadmap format:
## Q1 — [Theme]
Product deliverables: [list]
Platform / infra work: [list]
Tech debt addressed: [list]
Hiring: [roles]
Risk: [what could derail this quarter]
[repeat for Q2, Q3, Q4]
## Investment split (target)
- New product features: [X]%
- Platform and infrastructure: [X]%
- Tech debt reduction: [X]%
- Reliability and on-call: [X]%
Target ratio for healthy engineering orgs:
- Early stage: 70/15/10/5 (ship fast, worry about debt later)
- Growth stage: 60/20/15/5 (start investing in platform)
- Scale stage: 50/25/20/5 (debt and reliability become existential)
## Dependencies and blockers
What must happen outside engineering for this roadmap to succeed?
- Product decisions needed by [date]
- Design resources needed in [quarter]
- Data / legal / compliance approvals needed for [feature]
interface DORAMetrics {
// Deployment Frequency: how often you deploy to production
deploymentFrequency: 'multiple_per_day' | 'daily' | 'weekly' | 'monthly' | 'less_than_monthly'
// Lead Time for Changes: time from commit to production
leadTimeHours: number
// Change Failure Rate: % of deployments causing incidents
changeFailureRate: number // 0.0 to 1.0
// Mean Time to Restore: average time to recover from an incident
mttrHours: number
}
// Elite performer benchmarks (DORA State of DevOps report)
const DORA_ELITE: DORAMetrics = {
deploymentFrequency: 'multiple_per_day',
leadTimeHours: 1,
changeFailureRate: 0.05, // < 5%
mttrHours: 1,
}
// High performer benchmarks
const DORA_HIGH: DORAMetrics = {
deploymentFrequency: 'daily',
leadTimeHours: 24,
changeFailureRate: 0.10, // < 10%
mttrHours: 8,
}
// Use these to benchmark current state and set targets in engineering strategy
User: I'm a new CTO at a Series B startup (120 employees, 18 engineers). We have a Rails monolith that's 6 years old, 4 undifferentiated teams, 1 deploy per week, and 97% uptime. Board wants to 3x our product delivery in 12 months.
Output excerpt:
## Engineering Vision (12 months)
A 25-person engineering organisation shipping to production 10x per day with 99.5% uptime,
structured into 4 stream-aligned product teams and 1 platform team. A new engineer ships
to production in their first week.
## Current State Assessment
Architecture: Rails 6 monolith, 6 years of debt. Main limitation: feature flags don't exist,
so every deploy is all-or-nothing — this is why weekly deploys feel necessary.
Deployment velocity: 1 deploy/week = DORA "medium" performer. Target: daily minimum, multiple/day by Q4.
Reliability: 97% uptime = ~260 hours/year of downtime. Target: 99.5% = < 44 hours/year.
Team structure: 4 undifferentiated teams = everyone depends on everyone. Coordination overhead
explains why 18 engineers feel like they're moving slowly.
## Strategic Priorities
1. Feature flag infrastructure (Q1): Enable safe continuous deployment. Without this, velocity can't improve.
2. Platform team formation (Q1): 3 engineers off product work, full-time on CI/CD, observability, deployment
3. Team domain ownership (Q2): Assign clear product domain boundaries — stop the cross-team dependency web
4. Service extraction (Q3-Q4): Extract 2-3 highest-value bounded contexts from the monolith
## Build vs. Buy
| Capability | Recommendation | Rationale |
|---|---|---|
| Feature flags | Buy (LaunchDarkly) | Not a differentiator. $20k/year saves 8 eng-weeks |
| Observability | Buy (Datadog or Honeycomb) | Commodity. Buy now, build data pipeline later |
| CI/CD pipeline | Build on GitHub Actions | Already owned, team has expertise |
| Incident management | Buy (PagerDuty) | Solved problem, critical path |