🎯

database-driver-design

🎯Skill

from joannis/claude-skills

VibeIndex|
What it does

Designs robust Swift database client libraries with safe, type-safe protocols, connection management, and NIO integration for high-performance data access.

πŸ“¦

Part of

joannis/claude-skills(11 items)

database-driver-design

Installation

πŸ“‹ No install commands found in docs. Showing default command. Check GitHub for actual instructions.
Quick InstallInstall with npx
npx skills add joannis/claude-skills --skill database-driver-design
1Installs
-
AddedFeb 4, 2026

Skill Details

SKILL.md

'Expert guidance on building Swift database client libraries. Use when developers mention: (1) building a database driver, (2) wire protocol implementation, (3) connection pooling design, (4) state machines for protocol handling, (5) NIO channel handlers for databases, (6) backpressure in result streaming, (7) actor executor alignment with NIO.'

Overview

# Database Driver Design

This skill provides expert guidance on building production-quality database client libraries in Swift, covering wire protocol implementation, connection management, type-safe APIs, and integration with Swift Concurrency. Patterns are derived from exemplary implementations: valkey-swift and postgres-nio.

Agent Behavior Contract (Follow These Rules)

  1. Prefer parameterized queries - Never concatenate user input into SQL/command strings
  2. Use string interpolation for safety - Implement ExpressibleByStringInterpolation to convert values to bindings
  3. Design commands as types - Each command/query should be a struct with associated response type
  4. Implement state machines for protocols - Complex connection lifecycles need explicit state transitions
  5. Support backpressure - Row/result streaming must respect consumer demand
  6. Align actor executors - Use unownedExecutor to align with NIO event loops
  7. Pool connections properly - Implement keep-alive, idle timeout, and graceful shutdown

Core Patterns

Command as Type Pattern

Define commands as types with associated response types for compile-time safety:

```swift

public protocol DatabaseCommand: Sendable, Hashable {

associatedtype Response: Decodable

static var name: String { get }

func encode(into encoder: inout CommandEncoder)

}

public struct GET: DatabaseCommand {

public typealias Response = String?

public static var name: String { "GET" }

public let key: String

public func encode(into encoder: inout CommandEncoder) {

encoder.encode(Self.name, key)

}

}

```

SQL Injection Prevention via String Interpolation

```swift

public struct Query: ExpressibleByStringInterpolation {

public var sql: String

public var bindings: Bindings

public struct StringInterpolation: StringInterpolationProtocol {

var sql: String = ""

var bindings: Bindings = Bindings()

public mutating func appendLiteral(_ literal: String) {

sql.append(literal)

}

public mutating func appendInterpolation(_ value: T) {

bindings.append(value)

sql.append("$\(bindings.count)")

}

}

}

// Usage: let query: Query = "SELECT * FROM users WHERE id = \(userId)"

// Result: sql = "SELECT * FROM users WHERE id = $1", bindings = [userId]

```

Actor with NIO Executor Alignment

Eliminate context switches by aligning actor executor with NIO event loop:

```swift

public final actor Connection: Sendable {

nonisolated public let unownedExecutor: UnownedSerialExecutor

init(channel: any Channel) {

self.unownedExecutor = channel.eventLoop.executor.asUnownedSerialExecutor()

}

}

```

State Machine with Actions Pattern

Manage complex protocol state with explicit transitions and actions:

```swift

struct ConnectionStateMachine {

enum State {

case idle

case executing(QueryStateMachine)

case closing

case closed

case modifying // Prevents COW during mutations

}

enum Action {

case sendMessage(Message)

case forwardResult(Result)

case closeConnection

case none

}

private var state: State = .idle

mutating func handle(_ message: Message) -> Action {

switch (state, message) {

case (.idle, .query(let q)):

state = .executing(QueryStateMachine(q))

return .sendMessage(.parse(q))

// ... other transitions

}

}

}

```

Length-Prefixed Binary Encoding

Write length-prefixed data with placeholder and backfill:

```swift

extension Encodable {

func encodeRaw(into buffer: inout ByteBuffer) throws {

// Write placeholder for length (4 bytes for Int32)

let lengthIndex = buffer.writerIndex

buffer.writeInteger(Int32(0))

// Record position before encoding

let startIndex = buffer.writerIndex

// Encode the actual value

try self.encode(into: &buffer)

// Calculate and write actual length

let length = buffer.writerIndex - startIndex

buffer.setInteger(Int32(length), at: lengthIndex)

}

}

```

Protocol Hierarchy for Encoding/Decoding

Design tiered protocols for different encoding guarantees:

```swift

// Base: runtime-determined type, may throw

public protocol ThrowingDynamicTypeEncodable: Sendable {

func encode(into byteBuffer: inout ByteBuffer) throws

var dataType: DataType { get }

}

// Non-throwing variant

public protocol DynamicTypeEncodable: ThrowingDynamicTypeEncodable {

func encode(into byteBuffer: inout ByteBuffer)

}

// Static type info known at compile time

public protocol StaticTypeEncodable: ThrowingDynamicTypeEncodable {

static var dataType: DataType { get }

}

// Non-throwing + static type info (most efficient)

public protocol NonThrowingEncodable: StaticTypeEncodable, DynamicTypeEncodable {}

```

Variadic Generic Row Decoding

Decode multiple columns type-safely using parameter packs:

```swift

extension Row {

func decode(

_ types: (repeat each T).Type

) throws -> (repeat each T) {

var index = 0

return (repeat try decodeColumn((each T).self, at: &index))

}

}

// Usage: let (id, name, email) = try row.decode((Int.self, String.self, String.self))

```

Backpressure-Aware Streaming

Implement adaptive buffer strategy for result streaming:

```swift

struct AdaptiveBuffer: BackPressureStrategy {

var lowWatermark: Int

var highWatermark: Int

var currentTarget: Int

mutating func didYield(bufferDepth: Int) -> Bool {

// Shrink target if buffer too deep

if bufferDepth > currentTarget * 2 {

currentTarget = max(lowWatermark, currentTarget / 2)

}

return bufferDepth < currentTarget

}

mutating func didConsume(bufferDepth: Int) -> Bool {

// Grow target if buffer drains completely

if bufferDepth == 0 {

currentTarget = min(highWatermark, currentTarget * 2)

}

return bufferDepth < currentTarget

}

}

```

Connection Pool Integration

Conform connections to pool protocols:

```swift

extension Connection: PooledConnection {

public typealias ConnectionID = Int

}

struct KeepAliveBehavior: ConnectionKeepAliveBehavior {

typealias Connection = Connection

let frequency: Duration

func runKeepAlive(for connection: Connection) async throws {

_ = try await connection.ping()

}

}

final class ClientMetrics: ConnectionPoolObservabilityDelegate {

func connectionCreated(id: Int) { / metrics / }

func connectionLeased(id: Int) { / metrics / }

func connectionReleased(id: Int) { / metrics / }

func connectionClosed(id: Int) { / metrics / }

}

```

Depth-Limited Parsing

Prevent stack overflow with nested structures:

```swift

mutating func parseToken(maxDepth: Int = 100) throws -> Token {

guard maxDepth > 0 else {

throw ParsingError.tooDeeplyNested

}

switch tokenType {

case .array:

var elements: [Token] = []

for _ in 0..

elements.append(try parseToken(maxDepth: maxDepth - 1))

}

return .array(elements)

// ... other cases

}

}

```

Quick Decision Tree

  1. Implementing wire protocol encoding/decoding?

- Use length-prefixed messages with placeholder + backfill

- Define protocol tokens as enums with associated values

- Implement depth limits for nested structures

  1. Designing type-safe query API?

- Use ExpressibleByStringInterpolation for injection prevention

- Create Encodable/Decodable protocol hierarchies for type coercion

- Use variadic generics for multi-column row decoding

  1. Managing connections?

- Use actors with NIO executor alignment

- Implement hierarchical state machines

- Support cancellation via request IDs

  1. Implementing connection pooling?

- Conform to PooledConnection protocol

- Implement keep-alive behavior

- Track metrics via observability delegate

Triage-First Playbook

  • "SQL Injection vulnerability"

- Implement ExpressibleByStringInterpolation on query type

- Convert interpolated values to $1, $2... parameter placeholders

  • "Type mismatch when decoding results"

- Define Decodable protocol with typed throws

- Wrap errors with column/cell context

  • "Connection state corruption"

- Use hierarchical state machines with explicit transitions

- Add .modifying sentinel to prevent COW issues

  • "Backpressure not working"

- Implement adaptive buffer strategy

- Signal demand through data source protocol

Best Practices Summary

  1. Commands/Queries as types - Each operation is a struct with associated response type
  2. String interpolation for safety - Prevent injection by design
  3. Protocol hierarchies for encoding - Different guarantees (throwing, static types, etc.)
  4. Length-prefixed wire formats - Write placeholder, encode, backfill length
  5. Hierarchical state machines - Compose child machines for complex protocols
  6. Adaptive backpressure - Dynamically adjust buffer targets based on consumer rate
  7. Actor + NIO alignment - Eliminate context switches with unownedExecutor
  8. Cell/Row wrappers - Rich metadata for better error messages
  9. Prepared statement caching - Deduplicate concurrent preparations
  10. Graceful shutdown - Drain pending operations before closing

Key Libraries for Reference

  • valkey-swift (github.com/valkey-io/valkey-swift) - Excellent RESP3 protocol, cluster support, pub/sub
  • postgres-nio (github.com/vapor/postgres-nio) - Excellent query safety, type coercion, state machines

More from this repository10

🎯
hummingbird🎯Skill

I apologize, but I cannot generate a description without seeing the actual code or details about the "hummingbird" skill. Could you provide more context or share the code/repository details so I ca...

🎯
valkey🎯Skill

Provides Swift developers with a comprehensive Valkey/Redis client library for async key-value operations, cluster support, and advanced data structure management.

🎯
swift🎯Skill

Provides expert Swift development guidance, covering best practices, testing, logging, OpenTelemetry, platform-specific patterns, and memory-safe API design.

🎯
swift-library-design🎯Skill

Helps developers design Swift libraries with best practices, architectural guidance, and code generation recommendations.

🎯
swift-concurrency🎯Skill

Provides expert guidance on Swift Concurrency, helping developers implement safe async/await patterns, resolve thread safety issues, and navigate structured concurrency challenges.

🎯
postgres🎯Skill

Enables Swift developers to interact with PostgreSQL databases using async/await, type-safe queries, connection pooling, and prepared statements via the postgres-nio library.

🎯
wendy-contributing🎯Skill

Contributes code to a project by generating and suggesting improvements based on the current codebase and context.

🎯
swift-nio🎯Skill

Provides expert guidance on SwiftNIO networking, helping developers build performant, safe network applications with event-driven patterns and Swift Concurrency best practices.

🎯
linear🎯Skill

Manages Linear issue tracking via CLI, enabling listing, viewing, updating, and linking issues across different states and teams.

🎯
wendy🎯Skill

I apologize, but I cannot generate a description without seeing the actual code or context for the "wendy" skill. Could you provide more details about what the skill does, or share the relevant cod...