AgentSkillsCN

Skill: Database Architecture

技能：数据库架构

SKILL.md

Skill: Database Architecture

Name: database Description: Schema-driven database patterns with QueryBuilder Trigger: Database operations, schema changes, migrations, queries

Objective

All database operations use schema.rs and QueryBuilder - NO hardcoded SQL strings.

Architecture Overview

code
src/db/
  schema.rs           # Single source of truth for table/column definitions
  query_builder.rs    # Fluent API for SQL construction
  repository.rs       # Base repository pattern
  migrations.rs       # Version-controlled schema changes (uses QueryBuilder)
  *_repository.rs     # Entity-specific data access

Core Rules

  1. •Schema is Truth: All table/column definitions in schema.rs
  2. •QueryBuilder for Queries: Never write raw SQL strings
  3. •Repository Pattern: Each entity has its own repository
  4. •Migrations via Schema: Use table.create_table_sql(), never hardcode DDL

Correct Patterns

rust
// GOOD: Using QueryBuilder
let sql = QueryBuilder::select(&TOOLS_TABLE)
    .columns(&["id", "display_name"])
    .where_eq("enabled")
    .order_by("display_name", true)
    .build();

// GOOD: Schema-driven DDL
let ddl = TOOLS_TABLE.create_table_sql();

Anti-Patterns to Avoid

rust
// BAD: Hardcoded SQL
db.execute("SELECT * FROM tools WHERE id = ?", [id]).await?;

// BAD: Hardcoded DDL
db.execute("CREATE TABLE tools (id TEXT PRIMARY KEY)", ()).await?;

// BAD: SQL in migrations
db.execute("INSERT INTO schema_migrations...", params).await?;

Why This Matters

  • •Single point of maintenance for schema
  • •Compile-time verification of column names
  • •Consistent query patterns across codebase
  • •Easier migrations and schema evolution