tui-developer

作者: microsoft

生成具有聊天式界面的Go Bubble Tea TUI应用程序。当用户要求创建终端UI、聊天界面、交互式CLI、REPL等时使用。

npx skills add https://github.com/microsoft/ghqr --skill tui-developer

TUI Developer Skill

Generate minimal, idiomatic Go Bubble Tea TUI applications with chat-like interfaces using alt screen + internal viewport.

When to Use

Use this skill when the user asks to:

  • Create a terminal UI with Bubble Tea
  • Build a chat-like CLI interface
  • Implement a REPL or interactive terminal
  • Create a log viewer with input

Architecture

The TUI uses alt screen with an internal viewport for scrollable conversation history. Completed output is accumulated in a persisted string that is written to stdout after the alt screen exits, giving the user a transcript in their terminal scrollback.

┌───────────────────────────────┐
│  viewport (scrollable)        │  <- persisted history + live activity
│  ❯ user message               │
│  ● assistant response         │
│  ● Thinking... (live)         │
├───────────────────────────────┤
│  ─────────────────────────    │  <- separator (with top margin)
│  ❯ input bar                  │
│  ─────────────────────────    │
│  status bar          ↕ model  │  <- scroll arrow + model/mode
└───────────────────────────────┘

Key Design Decisions

  • tea.WithAltScreen() — full-screen TUI; avoids inline rendering artifacts
  • bubbles/viewport — scrollable content area; height = terminal height minus fixed chrome rows
  • persisted string — accumulates all finalized output; written to stdout on exit so the user keeps a transcript
  • buildViewportContent() — returns persisted + live activity (spinner/streaming) when a response is in progress
  • appendToViewport(content) — appends to persisted and calls updateViewport() (SetContent + GotoBottom)
  • No tea.WithMouseCellMotion() — mouse capture is omitted so the user can select text with the mouse
  • No tea.Println — all output goes through the viewport

Required Dependencies

import (
    tea "github.com/charmbracelet/bubbletea"
    "github.com/charmbracelet/bubbles/textinput"
    "github.com/charmbracelet/bubbles/spinner"
    "github.com/charmbracelet/bubbles/viewport"
    "github.com/charmbracelet/lipgloss"
)

Implementation Requirements

Model Structure

type model struct {
    // Dimensions
    width  int
    height int

    // Core components
    input    textinput.Model
    spinner  spinner.Model
    viewport viewport.Model

    // Content
    persisted string // finalized transcript (also written to stdout on exit)

    // State
    processing bool
    quitting   bool
    program    *tea.Program
}

Initialization

func newModel() *model {
    return &model{
        input:    newTextInput(),
        spinner:  newSpinner(),
        viewport: viewport.New(80, 20),
        width:    80,
    }
}

func (m *model) Init() tea.Cmd {
    return tea.Batch(textinput.Blink, m.spinner.Tick)
}

Viewport Height

Reserve 6 rows for chrome: two separators (each with a top margin counts as 2 rows) + input row + status row:

func (m *model) handleResize(msg tea.WindowSizeMsg) (tea.Model, tea.Cmd) {
    m.width = msg.Width
    m.height = msg.Height
    m.input.Width = msg.Width - 4
    m.viewport.Width = msg.Width
    vpHeight := msg.Height - 6
    if vpHeight < 5 {
        vpHeight = 5
    }
    m.viewport.Height = vpHeight
    m.viewport.SetContent(m.buildViewportContent())
    return m, nil
}

Viewport Content Helpers

// appendToViewport adds finalized content and scrolls to bottom.
func (m *model) appendToViewport(content string) {
    m.persisted += content
    m.updateViewport()
}

// buildViewportContent returns persisted history plus any live activity.
func (m *model) buildViewportContent() string {
    if m.processing {
        if live := m.renderActivity(); live != "" {
            return m.persisted + live
        }
    }
    return m.persisted
}

// updateViewport refreshes content and scrolls to bottom.
func (m *model) updateViewport() {
    m.viewport.SetContent(m.buildViewportContent())
    m.viewport.GotoBottom()
}

Keyboard Scrolling

case tea.KeyPgUp:
    m.viewport.HalfViewUp()
    return m, nil
case tea.KeyPgDown:
    m.viewport.HalfViewDown()
    return m, nil
case tea.KeyUp:
    m.viewport.LineUp(1)
    return m, nil
case tea.KeyDown:
    m.viewport.LineDown(1)
    return m, nil

Scroll Arrow in Status Bar

Display ↑/↓/↕ in the status bar to indicate scroll position:

func (m *model) scrollArrow() string {
    if m.viewport.YOffset == 0 && m.viewport.AtBottom() {
        return "" // all content fits
    }
    canUp := m.viewport.YOffset > 0
    canDown := !m.viewport.AtBottom()
    switch {
    case canUp && canDown:
        return "↕"
    case canUp:
        return "↑"
    default:
        return "↓"
    }
}

View Layout

View() renders the full screen: viewport + chrome. Return "" when quitting so the alt screen clears cleanly.

func (m *model) View() string {
    if m.quitting {
        return ""
    }
    separator := strings.Repeat("─", m.width-2)
    parts := []string{
        m.viewport.View(),
        separator,
        m.input.View(),
        separator,
        m.renderStatusBar(),
    }
    return strings.Join(parts, "\n")
}

Program Entry and Exit Transcript

After p.Run() returns, print persisted to stdout so the user gets a transcript in their shell:

func Run() error {
    m := newModel()
    p := tea.NewProgram(m, tea.WithAltScreen())
    m.program = p

    finalModel, err := p.Run()

    if fm, ok := finalModel.(*model); ok && fm.persisted != "" {
        fmt.Print(fm.persisted)
    }
    return err
}

Ctrl+C Behaviour

First Ctrl+C clears the input (with a 2-second warning); second Ctrl+C within that window exits:

func (m *model) handleCtrlC() (tea.Model, tea.Cmd) {
    now := time.Now()
    if m.ctrlCPressed && now.Sub(m.ctrlCTime) < 2*time.Second {
        m.quitting = true
        return m, tea.Quit
    }
    m.input.SetValue("")
    m.ctrlCPressed = true
    m.ctrlCTime = now
    return m, tea.Tick(2*time.Second, func(time.Time) tea.Msg {
        return ctrlCClearMsg{}
    })
}

Quality Checklist

Before returning generated code, verify:

  • Uses tea.WithAltScreen() — full-screen, no inline artifacts
  • Uses bubbles/viewport — scrollable content area
  • Does NOT use tea.Println — all output goes through the viewport
  • Does NOT use tea.WithMouseCellMotion() — preserves native text selection
  • persisted string accumulates all finalized content
  • appendToViewport = append to persisted + updateViewport()
  • buildViewportContent = persisted + live activity when processing
  • Viewport height = terminal height minus 6 fixed chrome rows
  • View() returns "" when quitting
  • After p.Run(), prints fm.persisted to stdout for transcript
  • PgUp/PgDn and Up/Down scroll the viewport
  • Scroll arrow (↑/↓/↕) shown in status bar
  • First Ctrl+C clears input; second Ctrl+C quits
  • Model uses pointer receiver (*model) for all methods
  • Update handlers are broken into focused methods

来自 microsoft 的更多技能

oss-growth
microsoft
OSS增长黑客角色
agent-framework-azure-ai-py
microsoft
使用Microsoft Agent Framework Python SDK(agent-framework-azure-ai)构建Azure AI Foundry代理。在创建使用AzureAIAgentsProvider的持久化代理、使用托管工具(代码解释器、文件搜索、网络搜索)、集成MCP服务器、管理对话线程或实现流式响应时使用。涵盖函数工具、结构化输出和多工具代理。
development
airunway-aks-setup
microsoft
在AKS上设置AI Runway——从裸集群到运行模型。涵盖集群验证、控制器安装、GPU评估、提供商设置和首次部署。适用场景:“设置AI Runway”、“接入AKS集群”、“安装AI Runway”、“airunway设置”、“将模型部署到AKS”、“在AKS上进行GPU推理”、“在AKS上配置KAITO”、“在AKS上运行LLM”、“在AKS上使用vLLM”、“在AKS上设置模型服务”、“AI Runway控制器”。
devops
appinsights-instrumentation
microsoft
使用Azure Application Insights对Web应用进行插桩的指南。提供遥测模式、SDK设置和配置参考。适用场景:如何对应用进行插桩、App Insights SDK、遥测模式、什么是App Insights、Application Insights指南、插桩示例、APM最佳实践。
devops
applicationinsights-web-ts
microsoft
使用Application Insights JavaScript SDK(@microsoft/applicationinsights-web)为浏览器/Web应用添加检测。用于真实用户监控(RUM)——页面视图、点击、AJAX/fetch依赖项、异常、自定义事件,以及与后端OpenTelemetry追踪关联的浏览器端GenAI代理追踪。涵盖SDK加载器脚本和npm设置、框架扩展(React、React Native、Angular)、点击分析、遥测初始化器,以及从浏览器发出的代理/工具/模型跨度所遵循的OTel GenAI语义约定。
devops
azure-ai-anomalydetector-java
microsoft
使用适用于 Java 的 Azure AI 异常检测器 SDK 构建异常检测应用程序。在实现单变量/多变量异常检测、时间序列分析或 AI 驱动的监控时使用。
development
azure-ai-language-conversations-py
microsoft
使用azure-ai-language-conversations Python SDK实现对话语言理解(CLU)。当使用ConversationAnalysisClient分析对话意图和实体、构建NLP功能或将语言理解集成到应用程序中时使用。
development
azure-ai-ml-py
microsoft
Azure Machine Learning SDK v2 for Python。用于机器学习工作区、作业、模型、数据集、计算资源和管道。 触发词:“azure-ai-ml”、“MLClient”、“工作区”、“模型注册表”、“训练作业”、“数据集”。
development