Build Go CLIs with spf13/cobra — command trees, flag parsing, arg validation, completions, and testable handlers.
Memory
golang-cli
Try itBuild, extend, and review Go CLIs with Cobra + Viper — flags, config, exit codes, signals, completion, testing.
What it does
An opinionated Go CLI reference for Cobra + Viper applications. Covers cmd/myapp/ layout, SilenceUsage/SilenceErrors on the root, persistent vs local flags, required-flag markers, Viper precedence (flags > env > config > defaults), ldflags version embedding, Unix exit codes, stdout/stderr discipline, signal.NotifyContext for graceful shutdown, generated shell completions, and CLI testing via cmd.OutOrStdout. Three modes: Build (new CLI), Extend (add to existing), Review (audit against the Common Mistakes checklist).
When to use it
- Building a new Cobra-based Go CLI from scratch
- Adding subcommands or persistent flags to an existing Go CLI
- Auditing a Go CLI for exit code, I/O, and config mistakes
- Wiring Viper so flags, env vars, and config files layer correctly
The skill document
Persona: You are a Go CLI engineer. You build tools that feel native to the Unix shell — composable, scriptable, and predictable under automation.
Modes:
- Build — creating a new CLI from scratch: follow the project structure, root command setup, flag binding, and version embedding sections sequentially.
- Extend — adding subcommands, flags, or completions to an existing CLI: read the current command tree first, then apply changes consistent with the existing structure.
- Review — auditing an existing CLI for correctness: check the Common Mistakes table, verify
SilenceUsage/SilenceErrors, flag-to-Viper binding, exit codes, and stdout/stderr discipline.
Go CLI Best Practices
Use Cobra + Viper as the default stack for Go CLI applications. Cobra provides the command/subcommand/flag structure and Viper handles configuration from files, environment variables, and flags with automatic layering. This combination powers kubectl, docker, gh, hugo, and most production Go CLIs.
When using Cobra or Viper, refer to the library's official documentation and code examples for current API signatures.
For trivial single-purpose tools with no subcommands and few flags, stdlib flag is sufficient.
Quick Reference
| Concern | Package / Tool |
|---|---|
| Commands & flags | github.com/spf13/cobra |
| Configuration | github.com/spf13/viper |
| Flag parsing | github.com/spf13/pflag (via Cobra) |
| Colored output | github.com/fatih/color |
| Table output | github.com/olekukonko/tablewriter |
| Interactive prompts | github.com/charmbracelet/bubbletea |
| Version injection | go build -ldflags |
| Distribution | goreleaser |
Project Structure
Organize CLI commands in cmd/myapp/ with one file per command. Keep main.go minimal — it only calls Execute().
myapp/
├── cmd/
│ └── myapp/
│ ├── main.go # package main, only calls Execute()
│ ├── root.go # Root command + Viper init
│ ├── serve.go # "serve" subcommand
│ ├── migrate.go # "migrate" subcommand
│ └── version.go # "version" subcommand
├── go.mod
└── go.sum
main.go should be minimal — see assets/examples/main.go.
Root Command Setup
The root command initializes Viper configuration and sets up global behavior via PersistentPreRunE. See assets/examples/root.go.
Key points:
SilenceUsage: trueMUST be set — prevents printing the full usage text on every errorSilenceErrors: trueMUST be set — lets you control error output format yourselfPersistentPreRunEruns before every subcommand, so config is always initialized- Logs go to stderr, output goes to stdout
Subcommands
Add subcommands by creating separate files in cmd/myapp/ and registering them in init(). See assets/examples/serve.go for a complete subcommand example including command groups.
Flags
See assets/examples/flags.go for all flag patterns:
Persistent vs Local
- Persistent flags are inherited by all subcommands (e.g.,
--config) - Local flags only apply to the command they're defined on (e.g.,
--port)
Required Flags
Use MarkFlagRequired, MarkFlagsMutuallyExclusive, and MarkFlagsOneRequired for flag constraints.
Flag Validation with RegisterFlagCompletionFunc
Provide completion suggestions for flag values.
Always Bind Flags to Viper
This ensures viper.GetInt("port") returns the flag value, env var MYAPP_PORT, or config file value — whichever has highest precedence.
Argument Validation
Cobra provides built-in validators for positional arguments. See assets/examples/args.go for both built-in and custom validation examples.
| Validator | Description |
|---|---|
cobra.NoArgs | Fails if any args provided |
cobra.ExactArgs(n) | Requires exactly n args |
cobra.MinimumNArgs(n) | Requires at least n args |
cobra.MaximumNArgs(n) | Allows at most n args |
cobra.RangeArgs(min, max) | Requires between min and max |
cobra.ExactValidArgs(n) | Exactly n args, must be in ValidArgs |
Configuration with Viper
Viper resolves configuration values in this order (highest to lowest precedence):
- CLI flags (explicit user input)
- Environment variables (deployment config)
- Config file (persistent settings)
- Defaults (set in code)
See assets/examples/config.go for complete Viper integration including struct unmarshaling and config file watching.
Example Config File (.myapp.yaml)
port: 8080
host: localhost
log-level: info
database:
dsn: postgres://localhost:5432/myapp
max-conn: 25
With the setup above, these are all equivalent:
- Flag:
--port 9090 - Env var:
MYAPP_PORT=9090 - Config file:
port: 9090
Version and Build Info
Version SHOULD be embedded at compile time using ldflags. See assets/examples/version.go for the version command and build instructions.
Exit Codes
Exit codes MUST follow Unix conventions:
| Code | Meaning | When to Use |
|---|---|---|
| 0 | Success | Operation completed normally |
| 1 | General error | Runtime failure |
| 2 | Usage error | Invalid flags or arguments |
| 64-78 | BSD sysexits | Specific error categories |
| 126 | Cannot execute | Permission denied |
| 127 | Command not found | Missing dependency |
| 128+N | Signal N | Terminated by signal (e.g., 130 = SIGINT) |
See assets/examples/exit_codes.go for a pattern mapping errors to exit codes.
I/O Patterns
See assets/examples/output.go for all I/O patterns:
- stdout vs stderr: NEVER write diagnostic output to stdout — stdout is for program output (pipeable), stderr for logs/errors/diagnostics
- Detecting pipe vs terminal: check
os.ModeCharDeviceon stdout - Machine-readable output: support
--outputflag for table/json/plain formats - Colors: use
fatih/colorwhich auto-disables when output is not a terminal
Signal Handling
Signal handling MUST use signal.NotifyContext to propagate cancellation through context. See assets/examples/signal.go for graceful HTTP server shutdown.
Shell Completions
Cobra generates completions for bash, zsh, fish, and PowerShell automatically. See assets/examples/completion.go for both the completion command and custom flag/argument completions.
Testing CLI Commands
Test commands by executing them programmatically and capturing output. See assets/examples/cli_test.go.
Use cmd.OutOrStdout() and cmd.ErrOrStderr() in commands (instead of os.Stdout / os.Stderr) so output can be captured in tests.
Common Mistakes
| Mistake | Fix |
|---|---|
Writing to os.Stdout directly | Tests can't capture output. Use cmd.OutOrStdout() which tests can redirect to a buffer |
Calling os.Exit() inside RunE | Cobra's error handling, deferred functions, and cleanup code never run. Return an error, let main() decide |
| Not binding flags to Viper | Flags won't be configurable via env/config. Call viper.BindPFlag for every configurable flag |
Missing viper.SetEnvPrefix | PORT collides with other tools. Use a prefix (MYAPP_PORT) to namespace env vars |
| Logging to stdout | Unix pipes chain stdout — logs corrupt the data stream for the next program. Logs go to stderr |
| Printing usage on every error | Full help text on every error is noise. Set SilenceUsage: true, save full usage for --help |
| Config file required | Users without a config file get a crash. Ignore viper.ConfigFileNotFoundError — config should be optional |
Not using PersistentPreRunE | Config initialization must happen before any subcommand. Use root's PersistentPreRunE |
| Hardcoded version string | Version gets out of sync with tags. Inject via ldflags at build time from git tags |
Not supporting --output format | Scripts can't parse human-readable output. Add JSON/table/plain for machine consumption |
Related Skills
See samber/cc-skills-golang@golang-project-layout, samber/cc-skills-golang@golang-dependency-injection, samber/cc-skills-golang@golang-testing, samber/cc-skills-golang@golang-design-patterns skills.
Related skills
Manage Go project dependencies with vetted workflows for go.mod, MVS, vulnerability scanning, and automated updates.
Measure Go performance with statistical rigor — write benchmarks, profile hot paths, and detect regressions before shipping.
Hand-judged Go style guidance on line breaks, declarations, control flow, and clarity beyond what linters enforce.
Golang semantic code intelligence via `gopls`, the official Go language server — go-to-definition, find references, call/implementation hierarchy, workspace symbol search, package API discovery, diagnostics, safe rename, refactors (extract/inline/fill/rewrite code actions), formatting, and generated tests. Reaches an agent via gopls's own MCP server (`go_*` tools), Claude Code's native `LSP` tool, or the `gopls` CLI. Use when navigating or refactoring Go code — jumping to a definition, finding call sites before a rename, understanding a file's or package's dependencies, running diagnostics after an edit, or extracting/inlining/renaming. Not for the published ecosystem — packages not in your `go.mod`, versions, licenses, importers — → See `samber/cc-skills-golang@golang-pkg-go-dev` skill (`godig`). Not for a whole-tree vulnerability audit → See `samber/cc-skills-golang@golang-security` skill (`govulncheck`).
Golang skills orchestrator — always active on any Golang coding, review, debug, or setup task. Reads the task context and loads the most relevant skills from samber/cc-skills-golang, often multiple at once: writing a gRPC service loads golang-grpc + golang-testing + golang-error-handling; debugging a panic loads golang-troubleshooting + golang-safety; auditing security loads golang-security + golang-lint + golang-safety. Also: disambiguates competing clusters when two skills seem to overlap (performance vs benchmark vs troubleshooting, samber/lo vs mo vs ro, DI cluster, safety vs security), and configures the project's agent-config file (CLAUDE.md, AGENTS.md, GEMINI.md, Cursor rules, or Copilot instructions) to force-trigger skills in a project (/golang-how-to configure).