Claude Code Plugins

Community-maintained marketplace

Feedback
1
0

This skill should be used when working with xterm.js terminal implementations, React-based terminal applications, WebSocket terminal communication, or refactoring terminal-related code. It provides battle-tested patterns, common pitfalls, and debugging strategies learned from building production terminal applications.

Install Skill

1Download skill
2Enable skills in Claude

Open claude.ai/settings/capabilities and find the "Skills" section

3Upload to Claude

Click "Upload skill" and select the downloaded ZIP file

Note: Please verify skill by going through its instructions before using it.

SKILL.md

name xterm-js
description This skill should be used when working with xterm.js terminal implementations, React-based terminal applications, WebSocket terminal communication, or refactoring terminal-related code. It provides battle-tested patterns, common pitfalls, and debugging strategies learned from building production terminal applications.

xterm.js Best Practices

Overview

This skill provides comprehensive best practices for building terminal applications with xterm.js, React, and WebSockets. It captures critical patterns discovered through debugging production terminal applications, including state management, WebSocket communication, React hooks integration, and terminal lifecycle management.

When to Use This Skill

Use this skill when:

  • Building or debugging xterm.js terminal implementations
  • Integrating xterm.js with React (hooks, state, refs)
  • Implementing WebSocket-based terminal I/O
  • Managing terminal persistence with tmux or similar backends
  • Refactoring terminal-related React components into custom hooks
  • Debugging terminal initialization, resize, or rendering issues
  • Implementing split terminal layouts or multi-window terminal management
  • Working on detach/reattach terminal functionality

Core Best Practices

1. Refs and State Management

Critical Pattern: Clear Refs When State Changes

Refs persist across state changes. When clearing state, also clear related refs.

// CORRECT - Clear both state AND ref
if (terminal.agentId) {
  clearProcessedAgentId(terminal.agentId)  // Clear ref
}
updateTerminal(id, { agentId: undefined })  // Clear state

Key Insight:

  • State (Zustand/Redux) = what the terminal is
  • Refs (useRef) = what we've processed
  • When state changes, check if related refs need updating

Common Scenario: Detach/reattach workflows where the same agentId returns from backend. Without clearing the ref, the frontend thinks it already processed this agentId and ignores reconnection messages.

See references/refs-state-patterns.md for detailed examples.

2. WebSocket Message Types

Critical Pattern: Know Your Destructive Operations

Backend WebSocket handlers often have different semantics for similar-looking message types:

  • type: 'disconnect' - Graceful disconnect, keep session alive
  • type: 'close' - FORCE CLOSE and KILL session (destructive!)
// WRONG - This KILLS the tmux session!
wsRef.current.send(JSON.stringify({
  type: 'close',
  terminalId: terminal.agentId,
}))

// CORRECT - For detach, use API endpoint only
await fetch(`/api/tmux/detach/${sessionName}`, { method: 'POST' })
// Don't send WebSocket message - let PTY disconnect naturally

Key Insight: Read backend code to understand what each message type does. "Close" often means "destroy" in WebSocket contexts.

See references/websocket-patterns.md for backend routing patterns.

3. React Hooks & Refactoring

Critical Pattern: Identify Shared Refs Before Extracting Hooks

When extracting custom hooks that manage shared resources:

// WRONG - Hook creates its own ref
export function useWebSocketManager(...) {
  const wsRef = useRef<WebSocket | null>(null)  // Creates NEW ref!
}

// RIGHT - Hook uses shared ref from parent
export function useWebSocketManager(
  wsRef: React.MutableRefObject<WebSocket | null>,  // Pass as parameter
  ...
) {
  // Uses parent's ref - all components share same WebSocket
}

Checklist Before Extracting Hooks:

  • Map out all refs (diagram which components use which refs)
  • Check if ref is used outside the hook
  • If ref is shared → pass as parameter, don't create internally
  • Test with real usage immediately after extraction

See references/react-hooks-patterns.md for refactoring workflows.

4. Terminal Initialization

Critical Pattern: xterm.js Requires Non-Zero Container Dimensions

xterm.js cannot initialize on containers with 0x0 dimensions. Use visibility-based hiding, not display:none.

// WRONG - Prevents xterm initialization
<div style={{ display: isActive ? 'block' : 'none' }}>
  <Terminal />
</div>

// CORRECT - All terminals get dimensions, use visibility to hide
<div style={{
  position: 'absolute',
  top: 0, left: 0, right: 0, bottom: 0,
  visibility: isActive ? 'visible' : 'hidden',
  zIndex: isActive ? 1 : 0,
}}>
  <Terminal />
</div>

Why This Works:

  • All terminals render with full dimensions (stacked via absolute positioning)
  • xterm.js can initialize properly on all terminals
  • visibility: hidden hides inactive terminals without removing dimensions
  • Use isSelected prop to trigger refresh when tab becomes active

Common Scenario: Tab-based terminal UI where switching tabs should show different terminals. After refresh, only active tab would render if using display: none.

5. useEffect Dependencies for Initialization

Critical Pattern: Early Returns Need Corresponding Dependencies

If a useEffect checks a ref and returns early, include ref.current in dependencies so it re-runs when ref becomes available.

// WRONG - Only runs once, may return early forever
useEffect(() => {
  if (!terminalRef.current) return  // Returns if null
  // Setup ResizeObserver
}, [])  // Never re-runs!

// CORRECT - Re-runs when ref becomes available
useEffect(() => {
  if (!terminalRef.current) return
  // Setup ResizeObserver
}, [terminalRef.current])  // Re-runs when ref changes!

Common Pattern: Wait for DOM refs AND library instances (xterm, fitAddon) before setup:

useEffect(() => {
  if (!terminalRef.current?.parentElement ||
      !xtermRef.current ||
      !fitAddonRef.current) {
    return  // Wait for all refs
  }
  // Setup ResizeObserver
}, [terminalRef.current, xtermRef.current, fitAddonRef.current])

6. Session Naming & Reconnection

Critical Pattern: Use Consistent Session Identifiers

When reconnecting, use the existing sessionName to find the existing PTY. Don't generate a new one.

// CORRECT - Reconnect to existing session
const config = {
  sessionName: terminal.sessionName,  // Use existing!
  resumable: true,
  useTmux: true,
}

// WRONG - Would create new session
const config = {
  sessionName: generateNewSessionName(),  // DON'T DO THIS
}

Key Insight: Tmux sessions have stable names. Use them as the source of truth for reconnection.

7. Multi-Window Terminal Management

Critical Pattern: Backend Output Routing Must Use Ownership Tracking

For multi-window setups, track which WebSocket connections own which terminals. Never broadcast terminal output to all clients.

// Backend: Track ownership
const terminalOwners = new Map()  // terminalId -> Set<WebSocket>

// On output: send ONLY to owners (no broadcast!)
terminalRegistry.on('output', (terminalId, data) => {
  const owners = terminalOwners.get(terminalId)
  owners.forEach(client => client.send(message))
})

Why: Broadcasting terminal output causes escape sequence corruption (DSR sequences) in wrong windows.

Frontend Pattern: Filter terminals by windowId before adding to agents:

// Check windowId BEFORE adding to webSocketAgents
if (existingTerminal) {
  const terminalWindow = existingTerminal.windowId || 'main'
  if (terminalWindow !== currentWindowId) {
    return  // Ignore terminals from other windows
  }
  // Now safe to add to webSocketAgents
}

See CLAUDE.md "Multi-Window Support - Critical Architecture" section for complete flow.

8. Testing Workflows

Critical Pattern: Test Real Usage Immediately After Refactoring

TypeScript compilation ≠ working code. Always test with real usage:

# After refactoring:
npm run build              # 1. Check TypeScript
# Open http://localhost:5173
# Spawn terminal            # 2. Test spawning
# Type in terminal          # 3. Test input (WebSocket)
# Resize window             # 4. Test resize (ResizeObserver)
# Spawn TUI tool            # 5. Test complex interactions

Refactoring Checklist:

  • TypeScript compilation succeeds
  • Spawn a terminal (test spawning logic)
  • Type in terminal (test WebSocket communication)
  • Resize window (test ResizeObserver)
  • Spawn TUI tool like htop (test complex ANSI sequences)
  • Check browser console for errors
  • Check backend logs
  • Run test suite: npm test

Prevention: Don't batch multiple hook extractions. Extract one, test, commit.

9. Debugging Patterns

Critical Pattern: Add Diagnostic Logging Before Fixing

When debugging complex state issues, add comprehensive logging first to understand the problem:

// BEFORE fixing, add logging:
console.log('[useWebSocketManager] 📨 Received terminal-spawned:', {
  agentId: message.data.id,
  requestId: message.requestId,
  sessionName: message.data.sessionName,
  pendingSpawnsSize: pendingSpawns.current.size
})

// Log each fallback attempt:
if (!existingTerminal) {
  existingTerminal = storedTerminals.find(t => t.requestId === message.requestId)
  console.log('[useWebSocketManager] 🔍 Checking by requestId:',
    existingTerminal ? 'FOUND' : 'NOT FOUND')
}

Benefits:

  • Shows exactly which code path is executing
  • Reveals data mismatches (wrong ID, missing state, etc.)
  • Helps users self-diagnose issues
  • Can be left in for production debugging

10. Multi-Step State Changes

Critical Pattern: Handle All Side Effects When Changing State

When a state change affects multiple systems, update all of them.

Checklist for Terminal State Changes:

  • Update Zustand state (terminal properties)
  • Clear/update refs (processedAgentIds, pending spawns)
  • Notify WebSocket (if needed)
  • Clean up event listeners
  • Update localStorage (if using persist)

Example (Detach):

// 1. API call
await fetch(`/api/tmux/detach/${sessionName}`, { method: 'POST' })

// 2. Clear ref (DON'T FORGET THIS!)
if (terminal.agentId) {
  clearProcessedAgentId(terminal.agentId)
}

// 3. Update state
updateTerminal(id, {
  status: 'detached',
  agentId: undefined,
})

Resources

references/

This skill includes detailed reference documentation organized by topic:

  • refs-state-patterns.md - Ref management patterns and examples
  • websocket-patterns.md - WebSocket communication and backend routing
  • react-hooks-patterns.md - React hooks refactoring workflows
  • testing-checklist.md - Comprehensive testing workflows
  • split-terminal-patterns.md - Split terminal and detach/reattach patterns
  • advanced-patterns.md - Advanced patterns (emoji width fix, mouse coordinate transformation, tmux reconnection)

Load these references as needed when working on specific aspects of terminal development.

Highlights from advanced-patterns.md:

  • Unicode11 Addon - Fix emoji/Unicode width issues (2 days of debugging → 1 line fix)
  • Mouse Coordinate Transformation - Handle CSS zoom/transform on terminal containers
  • Tmux Reconnection Best Practices - Prevent reconnecting to wrong sessions

scripts/

No scripts included - xterm.js integration is primarily about patterns and architecture, not executable utilities.

assets/

No assets included - this skill focuses on best practices and patterns rather than templates.