example-dev

name	example-dev
description	Development Standards skill template. Guides creation of project-specific development standards skills. Triggers on questions about tech stack, API conventions, code standards, coding style, or layered architecture.
model_tier	L1
skill_tier	atomic
version	1.2.0
status	active

不以规矩，不能成方圆。 ——《孟子·离娄上》 团队里最贵的不是服务器，是每个人对代码共识的每一次重建。

Development Standards Skill Template

Positioning: Atomic tier skill — provides an information lookup table for tech stack, layered architecture, and naming conventions. Orchestrated by change-model (functional), delegation (planning), and skill-builder-guide (meta).

Usage: AI scans project code → extracts standard patterns per layer → follows this template structure → generates a project-specific development standards skill.

Further reference: See references/code-scanning-guide.md for detailed code scanning methods.

Trigger Conditions

• Create a development standards skill / ask how to write tech stack or code standards
• Ask about coding style, naming conventions / layered code standard guidance
• Need standard pattern recognition methods / ask about dependency versions

Related Skills

• Code Map — file location methodology [L0]
• Change Model — change report generation
• Delegation — model tier delegation rules

---

1. Tech Stack

AI scans project dependency configuration files and extracts the actual tech stack in use.

Category	Technology	Version	Description
Language	{language}	{version}	Runtime environment
Framework	{framework}	{version}	Core framework
Database	{database}	{version}	Storage engine
Cache	{cache component}	{version}	Cache middleware
Message Queue	{MQ}	{version}	Async messaging

---

2. Code Layering Standards

2.1 Layered Architecture (General Model)

┌─────────────────────────────────────────────────────────┐
│  Interface Layer (API / Entry Layer)                     │
│  ├─ Request parameter validation & format conversion     │
│  ├─ Unified response wrapper                             │
│  └─ Error handling & exception translation               │
├─────────────────────────────────────────────────────────┤
│  Business Layer (Business Logic Layer)                   │
│  ├─ Core business logic                                  │
│  ├─ Transaction / consistency management                 │
│  ├─ Data assembly & transformation                       │
│  └─ External service orchestration                       │
├─────────────────────────────────────────────────────────┤
│  Data Layer (Data Access Layer)                          │
│  ├─ Persistence operations                               │
│  ├─ Query construction                                   │
│  └─ Cache operations                                     │
├─────────────────────────────────────────────────────────┤
│  Integration Layer (optional, per project reality)       │
│  ├─ Remote service calls                                 │
│  ├─ Message queue processing                             │
│  ├─ Scheduled tasks                                      │
│  └─ Third-party SDK wrappers                             │
└─────────────────────────────────────────────────────────┘

2.2 Layer Responsibility Boundaries

Layer	Responsibility	Prohibited
Interface Layer	Parameter validation, format conversion, unified response	Business logic, direct database access
Business Layer	Business logic, consistency management, data assembly	Direct request/response handling, inline query strings
Data Layer	Persistence operations, query construction	Business logic decisions
Integration Layer	Cross-service calls, async message processing	Core business logic implementation

2.3 Layer Identification Method

AI scans project files and records the actual layering situation.

Identification Item	Record Format
Interface layer directory	{directory name}
Interface layer naming	{naming pattern}
Business layer directory	{directory name}
Business layer naming	{naming pattern}
Data layer directory	{directory name}
Data layer naming	{naming pattern}
Integration layer directory	{directory name}

---

3. Standard Pattern Recognition

3.1 Interface Layer Standard Patterns

Check Item	Scan Target	Record Format
Route declaration	Routing configuration method	{annotation / config file / function}
Parameter validation	Validation approach	{declarative / manual / middleware}
Response wrapper	Unified response structure	{class name or function name}
Error handling	Error handling approach	{global / local}

Interface layer file structure:
Declare route & HTTP method → dependency-inject business components → each handler method:
    ├─ Parameter validation
    ├─ Call business layer
    └─ Return unified response format

3.2 Business Layer Standard Patterns

Check Item	Scan Target	Record Format
Interface definition	Whether abstract interface exists	{yes / no}
Consistency management	Transaction declaration method	{annotation / manual / framework-managed}
Parameter transformation	Model conversion approach	{method}
External calls	Cross-service call method	{SDK / RPC / message}

Business layer file structure:
Declare business component → each business method:
    ├─ Transform parameters to internal model
    ├─ Business logic processing
    ├─ Consistency boundary declaration
    ├─ Call data layer / integration layer
    └─ Return result model

3.3 Data Layer Standard Patterns

Check Item	Scan Target	Record Format
ORM method	Persistence approach	{ORM framework / raw queries / query builder}
Query organization	Query storage method	{annotation / separate file / inline}
Pagination method	Pagination strategy	{plugin / manual / framework built-in}

Data layer file structure:
Declare data access component → each data method:
    ├─ Build query conditions
    ├─ Parameterized query (injection prevention)
    ├─ Execute persistence operation
    └─ Return data model

3.4 Integration Layer Standard Patterns

Check Item	Scan Target	Record Format
Remote calls	Inter-service call method	{method}
Fault tolerance	Degradation / circuit-breaking strategy	{yes / no}
Timeout config	Timeout duration	{default / custom}
Message component	Message queue usage	{producer / consumer / both}

---

4. Naming Conventions

4.1 File Naming

AI extracts actual naming patterns from the project.

Type	Naming Pattern	Example
Interface layer files	{pattern}	{example}
Business layer files	{pattern}	{example}
Data layer files	{pattern}	{example}
Data model files	{pattern}	{example}

4.2 Method Naming

Operation Type	Naming Pattern	Example
Single query	{prefix} + {condition}	{example}
List query	{prefix} + {condition}	{example}
Paginated query	{prefix} + {condition}	{example}
Create	{prefix} + {target}	{example}
Update	{prefix} + {target}	{example}
Delete	{prefix} + {target}	{example}

4.3 Variable Naming

Type	Naming Rule
Constants	ALL_CAPS_UNDERSCORE / PascalCase
Member variables	camelCase / underscore-prefixed
Local variables	Semantically clear
Boolean variables	is / has / can prefix

---

5. Compatibility Pattern Recognition

5.1 What to Record

When scanning code, AI identifies patterns that differ from mainstream conventions but are preserved for historical reasons.

Scenario	Example	Record Notes
Legacy format	Old API response format differs from new API	Preserve, annotate applicable scope
Special optimization	High-performance handling for specific scenarios	Preserve, record the reason
Business exception	Unique transaction handling in a certain module	Preserve, annotate applicable conditions

See references/code-scanning-guide.md.

---

6. Coding Style Formation Process

① Project analysis → ② Code scanning (dispatch Haiku — L0) → ③ Style extraction
    → ④ Document generation → ⑤ User confirmation

AI execution steps:

1. Analyze project type and tech stack
2. Dispatch Haiku to scan code samples per layer (L0 task)
3. Extract standard patterns and compatibility patterns
4. Populate project-specific content following this template structure
5. Generate a project-specific development standards SKILL.md
6. User confirms or provides document adjustments

6.1 User Document Integration

If the user provides coding standard documents, follow the user's documents:

1. Read the coding standard documents provided by the user
2. Compare differences against scan results
3. Follow user documents as the authority, supplement with scan-discovered realities
4. Record discrepancies for user confirmation

See references/code-scanning-guide.md.

---

Handoff

After completing this skill, recommend the next skill based on output characteristics:

Condition	Recommend
New files created ≥3	{project}-code-map
Non-standard directory structure found	{project}-workflow

7. Model Tier

L1 — Sonnet / atomic tier: Contains reasoning work such as standard interpretation, code analysis, and style extraction. Scan subtasks are dispatched to L0 — Haiku.