Golang struct and interface design — composition, embedding, type assertions, type switches, DI via interfaces, field tags, pointer vs value receivers. Use when designing Go types, implementing interfaces, embedding, adding JSON/YAML/DB tags, or 'accept interfaces, return structs'.
Installer avec Codex ou Claude Copiez ce prompt, collez-le dans Codex, Claude ou un autre assistant, puis laissez-le vérifier la page du skill et l'installer pour vous.
Une commande directe contourne le prompt de vérification. Examinez la source avant de l'exécuter.
Golang struct and interface design — composition, embedding, type assertions, type switches, DI via interfaces, field tags, pointer vs value receivers. Use when designing Go types, implementing interfaces, embedding, adding JSON/YAML/DB tags, or 'accept interfaces, return structs'.
user-invocable
true
license
MIT
compatibility
Designed for Claude Code, Codex or similar harness, and for projects using Golang.
Persona: You are a Go type system designer. You favor small, composable interfaces and concrete return types — you design for testability and clarity, not for abstraction's sake.
Community default. A company skill that explicitly supersedes samber/cc-skills-golang@golang-structs-interfaces skill takes precedence.
Go Structs & Interfaces
Interface Design Principles
Keep Interfaces Small
"The bigger the interface, the weaker the abstraction." — Go Proverbs
Interfaces SHOULD have 1-3 methods. Small interfaces are easier to implement, mock, and compose. If you need a larger contract, compose it from small interfaces:
→ See samber/cc-skills-golang@golang-naming skill for interface naming conventions (method + "-er" suffix, canonical names)
type Reader interface {
Read(p []byte) (n int, err error)
}
type Writer interface {
Write(p []byte) (n int, err error)
}
// Composed from small interfacestype ReadWriter interface {
Reader
Writer
}
Compose larger interfaces from smaller ones:
type ReadWriteCloser interface {
io.Reader
io.Writer
io.Closer
}
Define Interfaces Where They're Consumed
Interfaces Belong to Consumers.
Interfaces MUST be defined where consumed, not where implemented. This keeps the consumer in control of the contract and avoids importing a package just for its interface.
// package notification — defines only what it needstype Sender interface {
Send(to, body string) error
}
type Service struct {
sender Sender
}
The email package exports a concrete struct — it doesn't need to know about .
Client
Sender
Accept Interfaces, Return Structs
Functions SHOULD accept interface parameters for flexibility and return concrete types for clarity. Callers get full access to the returned type's fields and methods; consumers upstream can still assign the result to an interface variable if needed.
// Good — accepts interface, returns concretefuncNewService(store UserStore) *Service { ... }
// BAD — NEVER return interfaces from constructorsfuncNewService(store UserStore) ServiceInterface { ... }
Don't Create Interfaces Prematurely
"Don't design with interfaces, discover them."
NEVER create interfaces prematurely — wait for 2+ implementations or a testability requirement. Premature interfaces add indirection without value. Start with concrete types; extract an interface when a second consumer or a test mock demands it.
// Bad — premature interface with a single implementationtype UserRepository interface {
FindByID(ctx context.Context, id string) (*User, error)
}
type userRepository struct { db *sql.DB }
// Good — start concrete, extract an interface later when neededtype UserRepository struct { db *sql.DB }
Make the Zero Value Useful
Design structs so they work without explicit initialization. A well-designed zero value reduces constructor boilerplate and prevents nil-related bugs:
// Good — zero value is ready to usevar buf bytes.Buffer
buf.WriteString("hello")
var mu sync.Mutex
mu.Lock()
// Bad — zero value is broken, requires constructortype Registry struct {
items map[string]Item // nil map, panics on write
}
// Good — lazy initialization guards the zero valuefunc(r *Registry) Register(name string, item Item) {
if r.items == nil {
r.items = make(map[string]Item)
}
r.items[name] = item
}
Avoid any / interface{} When a Specific Type Will Do
Since Go 1.18+, MUST prefer generics over any for type-safe operations. Use any only at true boundaries where the type is genuinely unknown (e.g., JSON decoding, reflection):
// Bad — loses type safetyfuncContains(slice []any, target any)bool { ... }
// Good — generic, type-safefuncContains[Tcomparable](slice []T, target T)bool { ... }
Key Standard Library Interfaces
Interface
Package
Method
Reader
io
Read(p []byte) (n int, err error)
Writer
io
Write(p []byte) (n int, err error)
Closer
io
Close() error
Stringer
fmt
String() string
error
builtin
Error() string
Handler
net/http
ServeHTTP(ResponseWriter, *Request)
Marshaler
encoding/json
MarshalJSON() ([]byte, error)
Unmarshaler
encoding/json
UnmarshalJSON([]byte) error
Canonical method signatures MUST be honored — if your type has a String() method, it must match fmt.Stringer. Don't invent ToString() or ReadData().
Compile-Time Interface Check
Verify a type implements an interface at compile time with a blank identifier assignment. Place it near the type definition:
var _ io.ReadWriter = (*MyBuffer)(nil)
This costs nothing at runtime. If MyBuffer ever stops satisfying io.ReadWriter, the build fails immediately.
Type Assertions & Type Switches
Safe Type Assertion
Type assertions MUST use the comma-ok form to avoid panics:
// Good — safe
s, ok := val.(string)
if !ok {
// handle
}
// Bad — panics if val is not a string
s := val.(string)
Type Switch
Discover the dynamic type of an interface value:
switch v := val.(type) {
casestring:
fmt.Println(v)
caseint:
fmt.Println(v * 2)
case io.Reader:
io.Copy(os.Stdout, v)
default:
fmt.Printf("unexpected type %T\n", v)
}
Optional Behavior with Type Assertions
Check if a value supports additional capabilities without requiring them upfront:
type Flusher interface {
Flush() error
}
funcwriteData(w io.Writer, data []byte)error {
if _, err := w.Write(data); err != nil {
return err
}
// Flush only if the writer supports itif f, ok := w.(Flusher); ok {
return f.Flush()
}
returnnil
}
This pattern is used extensively in the standard library (e.g., http.Flusher, io.ReaderFrom).
Struct & Interface Embedding
Struct Embedding
Embedding promotes the inner type's methods and fields to the outer type — composition, not inheritance:
type Logger struct {
*slog.Logger
}
type Server struct {
Logger
addr string
}
// s.Info(...) works — promoted from slog.Logger through Logger
s := Server{Logger: Logger{slog.Default()}, addr: ":8080"}
s.Info("starting", "addr", s.addr)
The receiver of promoted methods is the inner type, not the outer. The outer type can override by defining its own method with the same name.
When to Embed vs Named Field
Use
When
Embed
You want to promote the full API of the inner type — the outer type "is a" enhanced version
Named field
You only need the inner type internally — the outer type "has a" dependency
// Embed — Server exposes all http.Handler methodstype Server struct {
http.Handler
}
// Named field — Server uses the store but doesn't expose its methodstype Server struct {
store *DataStore
}
Dependency Injection via Interfaces
Accept dependencies as interfaces in constructors. This decouples components and makes testing straightforward:
type UserStore interface {
FindByID(ctx context.Context, id string) (*User, error)
}
type UserService struct {
store UserStore
}
funcNewUserService(store UserStore) *UserService {
return &UserService{store: store}
}
In tests, pass a mock or stub that satisfies UserStore — no real database needed.
Struct Field Tags
Use field tags for serialization control. Exported fields in serialized structs MUST have field tags:
type Order struct {
ID string`json:"id" db:"id"`
UserID string`json:"user_id" db:"user_id"`
Total float64`json:"total" db:"total"`
Items []Item `json:"items" db:"-"`
CreatedAt time.Time `json:"created_at" db:"created_at"`
DeletedAt time.Time `json:"-" db:"deleted_at"`
Internal string`json:"-" db:"-"`
}
Directive
Meaning
json:"name"
Field name in JSON output
json:"name,omitempty"
Omit field if zero value
json:"-"
Always exclude from JSON
json:",string"
Encode number/bool as JSON string
db:"column"
Database column mapping (sqlx, etc.)
yaml:"name"
YAML field name
xml:"name,attr"
XML attribute
validate:"required"
Struct validation (go-playground/validator)
Pointer vs Value Receivers
Use pointer (s *Server)
Use value (s Server)
Method modifies the receiver
Receiver is small and immutable
Receiver contains sync.Mutex or similar
Receiver is a basic type (int, string)
Receiver is a large struct
Method is a read-only accessor
Consistency: if any method uses a pointer, all should
Map and function values (already reference types)
Receiver type MUST be consistent across all methods of a type — if one method uses a pointer receiver, all methods should.
Preventing Struct Copies with noCopy
Some structs must never be copied after first use (e.g., those containing a mutex, a channel, or internal pointers). Embed a noCopy sentinel to make go vet catch accidental copies:
// noCopy may be added to structs which must not be copied after first use.// See https://pkg.go.dev/sync#noCopytype noCopy struct{}
func(*noCopy) Lock() {}
func(*noCopy) Unlock() {}
type ConnPool struct {
noCopy noCopy
mu sync.Mutex
conns []*Conn
}
go vet reports an error if a ConnPool value is copied (passed by value, assigned, etc.). This is the same technique the standard library uses for sync.WaitGroup, sync.Mutex, strings.Builder, and others.
Always pass these structs by pointer:
// Goodfuncprocess(pool *ConnPool) { ... }
// Bad — go vet will flag thisfuncprocess(pool ConnPool) { ... }
Cross-References
→ See samber/cc-skills-golang@golang-naming skill for interface naming conventions (Reader, Closer, Stringer)
→ See samber/cc-skills-golang@golang-design-patterns skill for functional options, constructors, and builder patterns
→ See samber/cc-skills-golang@golang-dependency-injection skill for DI patterns using interfaces
→ See samber/cc-skills-golang@golang-code-style skill for value vs pointer function parameters (distinct from receivers)
→ See samber/cc-skills-golang@golang-gopls skill for safe rename and the implementInterface code action — renaming a method or receiver that participates in interface satisfaction updates every call site and refuses a rename that would silently break the interface, which grep/sed cannot detect
Common Mistakes
Mistake
Fix
Large interfaces (5+ methods)
Split into focused 1-3 method interfaces, compose if needed
