К основному содержанию
скилл CX Codex

implementation-approach · shinpr/ai-coding-project-boilerplate — скилл для Claude Code: выбор стратегии

Скилл implementation-approach для Claude Code помогает выбрать способ реализации функции и оценить связанные риски. Используется при планировании изменений в существующем проекте.

MIT 1 установок ★ 240 обновлён 2026-11-13

Установить в Claude Code · Через Skills CLI (Vercel) Все способы и платформы ↓
npx skills add shinpr/ai-coding-project-boilerplate --skill implementation-approach -a claude-code
Открыть на GitHub

Что делает implementation-approach

Скилл анализирует архитектуру, потоки данных, зависимости и ограничения проекта, затем помогает выбрать стратегию: вертикальную, горизонтальную или гибридную.

Сначала скилл ищет минимальный дизайн, который выполняет требования, и проверяет, чем оправдано каждое добавление. Затем сопоставляет варианты реализации с ограничениями проекта и оценивает риски, для которых есть фактические основания. К примеру, вертикальная реализация ставит пользовательскую ценность на первое место, а подход с фокусом на фундамент — стабильность.

Выбор зависит от состояния репозитория и подтверждённых требований: скилл не подменяет их предположениями. Для установки в Claude Code нужна команда npx skills add shinpr/ai-coding-project-boilerplate --skill implementation-approach -a claude-code; файл навыка находится в ~/.claude/skills/implementation-approach/SKILL.md.

Какие задачи решает

  • выбрать стратегию реализации новой функции
  • сопоставить вертикальный, горизонтальный и гибридный подходы
  • найти минимальный дизайн, который выполняет требования
  • оценить риски изменений в существующей архитектуре

Примеры запросов агенту

  • Выбери стратегию реализации этой функции с учётом архитектуры проекта и рисков.
  • Сравни вертикальный и горизонтальный подходы для этого изменения.
  • Определи минимальный дизайн для задачи и объясни, какие риски требуют контроля.

Как установить implementation-approach

1Через Skills CLI (Vercel)

npx skills add shinpr/ai-coding-project-boilerplate --skill implementation-approach -a claude-code

2Вручную: положить файлы

~/.claude/skills/implementation-approach/SKILL.md

Папка скилла целиком; для одного проекта — .claude/skills/implementation-approach/

По шагам, с командами для каждой платформы: Как установить скилл в Claude Code: пять способов

Вопросы об implementation-approach

Как установить implementation-approach в Claude Code?

Выполните npx skills add shinpr/ai-coding-project-boilerplate --skill implementation-approach -a claude-code. Файл скилла будет доступен по пути ~/.claude/skills/implementation-approach/SKILL.md.

Что делает скилл implementation-approach?

Он помогает выбрать стратегию реализации функции, начиная с анализа проекта и минимального дизайна. Также скилл оценивает риски, связанные с выбранным подходом.

Какие стратегии реализации сравнивает скилл?

Он рассматривает вертикальный, горизонтальный и гибридный подходы, а также паттерны вроде Strangler и Adapter, если соответствующая проблема есть в репозитории.

Исходный текст

SKILL.md · MIT
---
name: implementation-approach
description: Selects implementation strategy (vertical slice, horizontal, or hybrid) with risk assessment. Use when planning feature implementation.
---

# Implementation Strategy Selection Framework (Meta-cognitive Approach)

## Meta-cognitive Strategy Selection Process

### Phase 1: Decision-Sufficient Current State Analysis

**Core Question**: "What does the existing implementation look like?"

#### Analysis Framework
```yaml
Architecture Analysis: Responsibility separation, data flow, dependencies, technical debt
Implementation Quality Assessment: Code quality, test coverage, performance, security
Historical Context Understanding: Current form rationale, past decision validity, constraint changes, requirement evolution
```

#### Meta-cognitive Question List
- What is the true responsibility of this implementation?
- Which parts are business essence and which derive from technical constraints?
- What dependencies or implicit preconditions are unclear from the code?
- What benefits and constraints does the current design bring?

Stop when the current-state facts that could change responsibility, reuse, option validity, total complexity, a contract, or verification have been checked; find existing code as coding-standards Change Boundary and Reference Representativeness describes.

**Completion evidence**: inspected paths, observed architecture/data-flow facts, known constraints, inferred historical rationale labeled as inferred, and unknowns that could change strategy selection.

**Transition**: proceed when every strategy-relevant claim is observed, explicitly inferred with evidence, or recorded as unknown.

### Phase 2: Design Convergence

**Core Question**: "What is the smallest design that delivers the current required outcome, and what evidence forces each addition beyond it?"

Complete these steps in order before exploring implementation strategies:

1. **Existing-Surface Baseline**: Form the simplest end-to-end path that delivers the current outcome through existing responsibilities. Explicit requirements and accepted decisions are binding; suggested mechanisms remain candidates.
2. **Evidence Check**: Test that path against current requirements, verified constraints, observed in-scope problems, and evidence-backed material risks. Keep only the unmet conditions that can change the selected design. Where the requirements leave a quality level unset, such as performance, freshness, consistency, availability, or protection beyond the required trust boundary, test against the lowest level that still delivers the outcome and record it as the selected level.
3. **Targeted Comparison**: For each unmet condition, test reuse, derivation from existing data, on-demand computation, or responsibility at the current caller or boundary before adding design surface. Compare viable choices by total complexity across the dimensions that materially differ: user decisions, settings, modes, concepts, outputs, persistent state, implementation paths, UX, runtime, implementation, testing, documentation, and maintenance. A second implementation of a responsibility that already has an owner with the same contract and change reason, including one added earlier in the same change, adds design surface; reusing or relocating that owner removes it. Select the lowest-total-complexity choice that satisfies the condition.
4. **Subtraction Check**: Remove each proposed addition and re-test its governing condition. Retain it only when the outcome, a required boundary, or necessary proof becomes unmet at its origin (a requirement, an accepted decision, or a repository rule or contract); a criterion, invariant, or constraint that a workflow artifact derived from that origin is narrowed to what the origin requires instead.

Candidate paths and rejected additions remain active analysis. The durable output is the **Selected Design**: the complete chosen path plus evidence for each added design surface and the condition that fails when it is removed. Only an accepted ADR may retain alternatives as decision history. During implementation, use the same convergence check without producing a separate artifact.

**Completion evidence**: one complete Selected Design; every added design surface names its current evidence, why lower-surface resolutions fail, and its subtraction result.

**Transition**: proceed when every supporting claim is observed, explicitly inferred with evidence, or recorded as unknown; route an unknown that blocks the next step as an exact evidence prerequisite. User interaction is required only when the unknown requires changing the outcome, desired-future requirements, or non-goals, or authorizing an irreversible action.

### Phase 3: Strategy Exploration and Creation

**Core Question**: "When determining before -> after, what implementation patterns or strategies should be referenced?"

#### Strategy Discovery Process
```yaml
Direct Strategy: The smallest repository-supported strategy that delivers the Selected Design within the hard constraints
Research Order: repository patterns first; then official documentation for the resolved dependency version; then maintained OSS implementations; use literature/blogs only as supplements labeled non-authoritative
```

#### Reference Strategy Patterns

Use a pattern only when its named migration or dependency problem exists in the repository.

**Legacy Handling Strategies**:
- Strangler Pattern: Gradual migration through phased replacement
- Facade Pattern: Complexity hiding through unified interface
- Adapter Pattern: Bridge with existing systems

**New Development Strategies**:
- Feature-driven Development: Vertical implementation prioritizing user value
- Foundation-driven Development: Foundation-first construction prioritizing stability
- Risk-driven Development: Prioritize addressing maximum risk elements

**Integration/Migration Strategies**:
- Proxy Pattern: Transparent feature extension
- Decorator Pattern: Phased enhancement of existing features
- Bridge Pattern: Flexibility through abstraction

**Completion evidence**: the direct strategy mapped to the observed constraints it satisfies and the constraints it leaves unresolved, plus an alternative only when it would materially change risk, compatibility, rollout, or the early verification point.

**Transition**: proceed when the direct strategy and any retained alternative are comparable against the same constraint set.

### Phase 4: Risk Assessment and Control

**Core Question**: "What risks arise when applying this to the existing implementation, and which control measurably reduces likelihood or impact while preserving verification and rollback?"

Evaluate only risk categories for which current evidence can change the strategy, public contract, rollout, rollback, or verification boundary.

#### Risk Analysis Matrix
```yaml
Technical Risks: System impact, data consistency, performance degradation, integration complexity
Operational Risks: Service availability, deployment downtime, process changes, rollback procedures
```

**Completion evidence**: each evidenced material risk has likelihood/impact evidence and one preventive or containment control.

**Transition**: proceed when every high-impact risk has either a control or a blocking escalation.

### Phase 5: Constraint Compatibility Verification

**Core Question**: "What are this project's constraints?"

Check only constraints evidenced by the governing requirements, repository, external contracts, or current environment that can change the selected strategy or verification boundary.

#### Constraint Checklist
```yaml
Technical Constraints: Library compatibility, resource capacity, mandatory requirements, numerical targets
Temporal Constraints: Deadlines/priorities, dependencies, milestones, learning periods
```

**Completion evidence**: each constraint is observed, inferred, or unknown; every unknown that can invalidate a candidate names the exact evidence prerequisite.

**Transition**: proceed when remaining unknowns cannot change the valid candidate set, or the user resolves them.

### Phase 6: Implementation Approach Decision

Select the approach that satisfies all hard constraints and current requirements with the lowest transition risk and smallest verification delay. Use lifecycle cost and implementation effort only as tiebreakers after requirement coverage, compatibility, and risk control are equal.

#### Vertical Slice (Feature-driven)
**Characteristics**: Vertical implementation across all layers by feature unit
**Application Conditions**: Low inter-feature dependencies, output in user-usable form, changes needed across all architecture layers
**Verification Method**: End-user value delivery at each feature completion

#### Horizontal Slice (Foundation-driven)
**Characteristics**: Phased construction by architecture layer
**Application Conditions**: Foundation system stability important, multiple features depend on common foundation, layer-by-layer verification effective
**Verification Method**: Integrated operation verification when all foundation layers complete

#### Hybrid (Creative Combination)
**Characteristics**: Flexible combination according to project characteristics
**Application Conditions**: Unclear requirements, need to change approach per phase, transition from prototyping to full implementation
**Verification Method**: Assign L1 when the phase produces end-user-operable behavior, L2 when it produces a testable internal behavior or contract, and L3 only when the phase produces build-time structure with no runnable behavior yet

For Hybrid, assign one explicit L1/L2/L3 verification level and observable completion result to every phase.

**Completion evidence**: one selected approach, its phase boundaries, integration points, and a verification result for every phase.

**Transition**: proceed to documentation when the selected approach covers every hard constraint and its risks have controls. Otherwise return to candidate exploration (Phase 3), or to Design Convergence (Phase 2) when a Phase 4-5 result changes the Selected Design or its evidence.

### Phase 7: Decision Rationale Documentation

Return the following structure in the Design Doc or planning handoff:

```yaml
implementationApproachDecision:
  observedConstraints: [<constraint + evidence>]
  inferredConstraints: [<constraint + evidence and inference>]
  unknowns: [<unknown + required evidence or decision>]
  selectedApproach: <vertical | horizontal | hybrid description>
  selectionRationale: <hard-constraint coverage, compatibility, risk control, and total-complexity basis>
  addedDesignSurface: [<addition + current evidence + lower-surface insufficiency + subtraction result>]
  phaseVerification: [<phase + L1/L2/L3 + observable completion evidence>]
```

Candidate approaches and rejection reasoning remain active analysis unless an accepted ADR owns them as decision history.

**Completion evidence**: the selected approach and every added design surface trace to an observed constraint, accepted inference, or resolved value-boundary decision.

## Verification Level Definitions

Priority for completion verification of each task:

- **L1: Functional Operation Verification** - Operates as an end-user feature (e.g., a user can execute a search and receive results)
- **L2: Test Operation Verification** - Tests that prove the changed behavior pass (e.g., type definition tests)
- **L3: Build Success Verification** - No compile errors (e.g., interface definitions)

**Priority**: L1 > L2 > L3 in order of verifiability importance

## Integration Point Definitions

Define integration points according to selected strategy:
- **Strangler-based**: When switching between old and new systems for each feature
- **Feature-driven**: When users can actually use the feature
- **Foundation-driven**: When all architecture layers are ready and E2E tests pass
- **Hybrid**: When individual goals defined for each phase are achieved

## Decision Gate Checklist

- [ ] Phase 1 evidence exists before strategy selection
- [ ] Phase 2 produces one complete Selected Design and every added design surface maps to current evidence, lower-surface insufficiency, and a subtraction result that leaves its origin unmet
- [ ] Candidate generation includes combinations when no listed strategy satisfies all hard constraints
- [ ] Every evidenced material risk has a control
- [ ] Every hard constraint maps to the selected approach
- [ ] Phase 7 output records the selection, total-complexity basis, and added design surface; alternatives appear only in an accepted ADR

When evidence required by a checked item is unknown, stop at that phase and report the exact repository evidence prerequisite. User interaction is required only when the unknown requires changing the outcome, desired-future requirements, or non-goals, or authorizing an irreversible action.

## Guidelines for Meta-cognitive Execution

1. **Leverage Known Patterns**: Use a pattern when its problem and trade-off match the observed repository state
2. **Evidence-Ordered Research**: Use repository evidence, version-matched official documentation, maintained OSS examples, then supplementary secondary sources
3. **Apply 5 Whys**: Pursue root causes to grasp essence
4. **Multi-perspective Evaluation**: Complete the evidence and transition checks for Phases 1-5
5. **Strategy Composition**: Combine strategies when one strategy cannot satisfy all hard constraints
6. **Decision Traceability**: Map every selection reason to evidence in the Phase 7 output

© shinpr · MIT · shinpr/ai-coding-project-boilerplate

Похожие скиллы

Все скиллы по теме «Настройка агента: память, контекст, субагенты» →

Сначала — implementation-approach от других авторов.

implementation-approach shinpr Скилл implementation-approach для Claude Code помогает выбрать стратегию реализации и критерии проверки. Он включается при планировании изменений и сравнении вариантов. CX Codex +1 ★ 720звёзд репо notion-spec-to-implementation makenotion Скилл notion-spec-to-implementation для Claude Code превращает спецификацию Notion в план работ и задачи. Он сверяет реализацию с требованиями и фиксирует прогресс. CX Codex +1 389установок implementation-planner galdawave Скилл implementation-planner превращает требования в поэтапный технический план. Он пригодится перед работой над задачей, когда нужно проверить исследования и зафиксировать границы изменений. CX Codex +1 ★ 452звёзд репо 10x-implement przeprogramowani Команда 10x-implement для пошагового внедрения утверждённых технических планов прямо в Claude Code. Читает план, отслеживает прогресс и проверяет каждый этап перед переходом к следующему. CX Codex +1 ★ 267звёзд репо gepetto softaworks gepetto превращает Markdown-спецификацию в план реализации: изучает тему, уточняет требования у пользователя и отправляет план на внешнюю проверку. Запускайте его до начала сложной разработки. CX Codex +1 1 600установок update-implementation-plan официальный github Скилл update-implementation-plan для GitHub Copilot обновляет план реализации под новые требования, рефакторинг, обновление пакетов и изменения архитектуры. Превращает план в исполнимый список задач. CX Codex +1 11 107установок planning samhvw8 Скилл planning изучает кодовую базу и помогает оформить технический план с архитектурой, этапами и разбором задач. Пригодится перед разработкой функции или изменением системы. CX Codex +1 109установок

Спонсоры проекта

Компании, которые помогают держать аналитику открытой