code-quality

Code Quality Standards

Apply these standards when writing or editing code.

Core Principles

No Hardcoded Values

Language	Bad	Good
TypeScript	const timeout = 5000;	const timeout = config.timeout;
Go	url := "https://api.example.com"	url := cfg.APIURL
Rust	let port = 8080;	let port = config.port;
C#	var connStr = "Server=localhost";	var connStr = configuration["ConnectionString"];
Java	int timeout = 5000;	int timeout = appConfig.getTimeout();
Swift	let apiUrl = "https://api.example.com"	let apiUrl = Configuration.apiURL
Kotlin	val timeout = 5000	val timeout = BuildConfig.TIMEOUT
Elixir	@timeout 5000	@timeout Application.compile_env(:app, :timeout)

Proper Error Handling

Errors must be logged with context and propagated meaningfully. Never swallow errors silently.

Language	Pattern
TypeScript	try { ... } catch (error) { logger.error('context', { error }); throw new AppError('message', { cause: error }); }
Go	if err != nil { return fmt.Errorf("functionName: %w", err) } — NEVER bare return err
Rust	operation().context("what failed")? or operation().map_err(|e| AppError::new(e))?
C#	catch (Exception ex) { logger.LogError(ex, "context"); throw new AppException("message", ex); }
Java	catch (Exception e) { log.error("context", e); throw new AppException("message", e); }
Swift	do { try operation() } catch { logger.error("context: \(error)"); throw AppError.operationFailed(cause: error) }
Kotlin	runCatching { operation() }.onFailure { e -> logger.error("context", e); throw AppException("message", e) }
Elixir	case operation() do {:error, reason} -> Logger.error("context", reason: reason); {:error, {:wrapped, reason}} end

Type Safety

• Type hints on public methods
• Interfaces/traits for data structures
• Avoid escape hatches (any in TS, interface{} in Go, Object in Java)

Naming Conventions

• Descriptive, intention-revealing names
• Follow language idiom: camelCase (TS/Java), snake_case (Rust/Elixir), MixedCaps (Go), PascalCase (C#)
• No abbreviations unless universal (URL, ID, etc.)

Language-Specific Anti-Patterns

Go

Anti-Pattern	Fix
Bare return err	return fmt.Errorf("context: %w", err)
Manual loops for slices operations (Go 1.21+)	Use slices.Contains, slices.Sort, etc.
interface{} (Go 1.18+)	Use any
if a > b { return a } (Go 1.21+)	Use max(a, b)
Unbounded goroutines	Use errgroup.Group for bounded concurrency
Missing context.Context	Always pass as first parameter

Rust

Anti-Pattern	Fix
.unwrap() in library code	Use ? operator with proper error types
.clone() to escape borrow checker	Restructure ownership or use references
Manual error types without thiserror	Use thiserror for library, anyhow for apps
Blocking in async context	Use tokio::spawn_blocking

TypeScript

Anti-Pattern	Fix
any type	Use proper types or unknown with type guards
@ts-ignore / @ts-expect-error	Fix the type error properly
Not running tsc --noEmit	Always verify types compile
== null without strictNullChecks	Enable strict mode in tsconfig

C#

Anti-Pattern	Fix
new HttpClient() per request	Use IHttpClientFactory
Startup.cs on .NET 6+	Use minimal hosting with WebApplication.CreateBuilder
Task.Result / Task.Wait()	Use await — blocking on async causes deadlocks
Nullable warnings ignored	Enable <Nullable>enable</Nullable>

Java

Anti-Pattern	Fix
JPA entities in WebFlux pipelines	Use R2DBC for reactive data access
javax.* on Jakarta EE 10+	Use jakarta.* namespace
Raw @Autowired field injection	Use constructor injection
Catching Exception broadly	Catch specific exception types

Swift

Anti-Pattern	Fix
ObservableObject / @Published on iOS 17+	Use @Observable macro
NavigationView	Use NavigationStack / NavigationSplitView
foregroundColor()	Use foregroundStyle()
DispatchQueue.main.async	Use @MainActor
Force unwrap (!) outside tests	Use guard let or if let
Modifying .pbxproj directly	Create files, add to Xcode manually

Kotlin (Android)

Anti-Pattern	Fix
GlobalScope for coroutines	Use viewModelScope or lifecycleScope
LiveData in Compose code	Use StateFlow + collectAsStateWithLifecycle()
kapt for annotation processing	Migrate to KSP
Blocking calls on Main dispatcher	Use withContext(Dispatchers.IO)
Allocations inside composable functions	Use remember {}
MaterialTheme.colors (Material2)	Use MaterialTheme.colorScheme (Material3)

Elixir

Anti-Pattern	Fix
try/catch for flow control	Use tagged tuples {:ok, _} / {:error, _} and with chains
String.to_atom/1 with user input	Use String.to_existing_atom/1 — atoms are not garbage collected
Blocking handle_call with long operations	Offload to Task.Supervisor, handle result in handle_info/2
Missing handle_info/2 catch-all	Add def handle_info(_msg, state), do: {:noreply, state}
Hardcoded config values	Use Application.compile_env(:app, :key)
Writing macros when functions suffice	Use regular functions — macros are hard to debug and AI-generated macros are frequently incorrect

Error Handling Checklist

• External calls wrapped in error handling
• Meaningful error messages with context
• Errors logged with structured data
• Graceful degradation where appropriate
• Fail fast on invalid state

Security Considerations

• No sensitive data in logs
• Input validation at boundaries
• No injection vulnerabilities (SQL, command, etc.)
• Authentication/authorization checks
• Secure defaults

Instrumentation

Add observability with structured logging at boundaries:

Language	Example
TypeScript	logger.info('operation completed', { operation: 'createUser', userId, duration })
Go	slog.Info("operation completed", "operation", "createUser", "userId", userId, "duration", duration)
Rust	tracing::info!(operation = "create_user", user_id = %id, duration = ?elapsed, "operation completed")
C#	logger.LogInformation("Operation {Op} completed for {UserId} in {Duration}ms", op, userId, duration)
Java	log.info("Operation {} completed for userId={} in {}ms", op, userId, duration)
Swift	Logger().info("Operation \(op) completed for userId=\(userId) in \(duration)ms")
Kotlin	Timber.i("Operation %s completed for userId=%s in %dms", op, userId, duration)
Elixir	Logger.info("operation completed", operation: :create_user, user_id: id, duration_ms: elapsed)

Logging levels:

• DEBUG: Detailed flow
• INFO: Key events
• WARNING: Recoverable issues
• ERROR: Failures requiring attention

When Reviewing Your Code

Before finishing, verify:

1. No hardcoded values
2. Error handling complete (language-idiomatic)
3. Types defined (no escape hatches)
4. Edge cases handled
5. Logging in place