Skip to content
Development
Skill

/nw-ux-tui-patterns

Terminal UI and CLI design patterns for product owners. Load when designing command-line tools, interactive terminal applications, or writing CLI-specific acceptance criteria.

From plugin
nwave
591200 skills34 agents27 commands
Install
$ npx -y skills add nWave-ai/nWave --skill nw-ux-tui-patterns --agent claude-code

How it fires

How this skill gets triggered: by you, by Claude, or both.

  • Fires itselfAuto-invocation. Claude auto-loads it when your prompt matches the work.Auto-invocation is when the right skill fires by itself at the right moment, driven by a FLOW.md router and a hook, instead of you invoking it by name. It is the difference between a skill being installed and a skill actually getting used.Read the full definition →
  • You can call itInvoke it directly when you want it.
  • Slash command/nw-ux-tui-patterns

Context preview

The summary Claude sees to decide when to auto-load this skill.

Terminal UI and CLI design patterns for product owners. Load when designing command-line tools, interactive terminal applications, or writing CLI-specific acceptance criteria.

SKILL.md

nw-ux-tui-patterns.SKILL.md
name: nw-ux-tui-patterns
description: Terminal UI and CLI design patterns for product owners. Load when designing command-line tools, interactive terminal applications, or writing CLI-specific acceptance criteria.
user-invocable: false
disable-model-invocation: true

TUI and CLI Patterns

Actionable terminal interface patterns for requirements gathering and design review. Use when target is CLI tool or interactive terminal application.

CLI Argument Design

Command Structure

Follow `program subcommand [flags] [arguments]` (e.g., `git commit -m "message"`).

Flags and Options

  • Both short (`-h`) and long (`--help`) for common flags
  • Reserve single-letter flags for frequently used options
  • Standard flags always: `--help`, `--version`, `--verbose`/`--quiet`, `--no-color`
  • Never accept secrets via flags (use files, stdin, or env vars)
  • Make flag order independent

Subcommands

  • Verbs for actions: `create`, `delete`, `list`, `show`, `update`
  • Nouns for targets: `user`, `project`, `config`
  • Pick one pattern (`tool resource action` or `tool action resource`) consistently
  • Provide shell completion scripts for discoverability

Argument Design Principles

Required arguments are positional; optional use flags. Accept stdin for piping. Support glob patterns for file arguments. Provide `--dry-run` for destructive/complex operations.

Interactive TUI Patterns

Framework Architectures

| Framework | Language | Architecture | Best For | |-----------|----------|-------------|----------| | **Bubble Tea** | Go | Elm (Model-Update-View) | Full TUI apps with complex state | | **Rich** | Python | Declarative rendering | Beautiful output, progress, tables | | **Ink** | JavaScript | React component model | Teams with React experience | | **Textual** | Python | CSS-like styling, widgets | Dashboard-style applications | | **Ratatui** | Rust | Immediate-mode rendering | High-performance terminal apps |

Key Architectural Pattern (Elm Architecture)

  • **Model**: Application state (single data structure)
  • **Update**: Pure function taking model + message, returns new model
  • **View**: Pure function rendering model to string

Unidirectional data flow makes TUI state predictable and testable.

Interactive Selection Patterns

Arrow keys to navigate, Enter to select. Type-ahead filtering for long lists. Multi-select with space bar, confirm with Enter. Show selected count with clear visual indicator. Support Esc to cancel.

Color and Formatting

Color Conventions

**Red**: errors, failures, deletions | **Yellow/amber**: warnings, attention | **Green**: success, confirmations, additions | **Blue/cyan**: information, links, highlights | **Dim/gray**: secondary info, metadata

Color Rules

  • Never use color as only way to convey information (accessibility)
  • Respect `NO_COLOR` environment variable (no-color.org)
  • Detect TTY: disable color when piped or redirected
  • Test with color-blindness simulators (protanopia, deuteranopia)

Formatting Patterns

Bold for emphasis and headings | Tables for structured data | Indentation for hierarchy | Horizontal rules for section breaks | Unicode box-drawing for visual structure (provide ASCII fallback)

Error Message Design

Every CLI error answers three questions:

1. **What happened?** Clear, jargon-free description 2. **Why?** The cause or context 3. **What to do?** Concrete next step

Good Example

Error: Could not connect to database at localhost:5432

  The connection was refused. The database server may not be
  running or is not accepting connections on this port.

  Try:
    1. Check if PostgreSQL is running:  pg_isready -h localhost -p 5432
    2. Start the server:                sudo systemctl start postgresql
    3. Verify the port in config:       cat ~/.config/myapp/database.yaml

  Documentation: https://docs.myapp.com/troubleshooting/database

Bad Example

Error: ECONNREFUSED

Error Guidelines

  • Never blame user ("You entered invalid..." becomes "The value ... is not valid because...")
  • Suggest most likely correction (did-you-mean for typos)
  • Exit codes consistent: 0=success, 1=general error, 2=usage error
  • Log verbose diagnostics to file, not stderr by default

Help Text Design

`--help` Output Structure

tool-name - One-line description

USAGE
  tool-name <command> [flags]

COMMANDS
  create    Create a new resource
  list      List existing resources
  delete    Remove a resource

FLAGS
  -h, --help       Show this help message
  -v, --version    Show version information
  -q, --quiet      Suppress non-error output

EXAMPLES
  # Create a new project
  tool-name create my-project --template=web

  # List all projects
  tool-name list --format=table

LEARN MORE
  Documentation:  https://docs.tool-name.com
  Report issues:  https://github.com/org/tool-name/issues

Help Text Principles

Lead with examples (users scan for examples first). Show common commands/flags first. One line per description. Link to detailed docs. Support both `tool help <cmd>` and `tool <cmd> --help`.

Output Design

Human-Readable (Default)

Detect TTY: terminal gets colors, tables, progress bars. Disable animations/spinners in non-TTY mode. Use structured formatting (headers, indentation, alignment).

Machine-Parseable Alternatives

  • `--json` for structured output (most important machine format)
  • `--plain` or `--tsv` for tabular data compatible with grep, awk, cut
  • `--quiet` for scripts needing only exit code

Output Contract

Once `--json` schema published, treat as API contract. Breaking changes require major version bump or `--output-version` flag.

Progress and Responsiveness

  • Respond within 100ms: output something immediately, especially before network calls
  • Spinner for operations 1-5 seconds
  • Progress bar with percentage for operations >5 seconds
  • Multi-step: `[3/7] Installing dependencies...`
  • Make operations idempotent where possible (safe to re-run after i
Read more
Ships withnwave

AI agents that guide you from idea to working code, with human judgment at every gate. nWave runs inside Claude Code. It breaks feature delivery into seven waves (discover, diverge, discuss, design, devops, distill, deliver).

Get the whole plugin