本文へ移動
cccskills
無料GitHub で公開

terminal-title

This skill should be used to update terminal window title with context. Triggers automatically at session start via hook. Also triggers on topic changes during conversation (debugging to docs, frontend to backend). Updates title with emoji + project + current topic.

インストール方法を見る

含まれるファイル(4)

  • SKILL.md9.8 KB
  • scripts/set_title.ps11.5 KB
  • scripts/set_title.sh1.4 KB
  • skill-report.json71.7 KB

SKILL.md(原文)

インストールする前に、エージェントに与えられる指示の中身を確認できます。

<!-- ABOUTME: Terminal title skill that automatically updates terminal window title --> <!-- ABOUTME: Uses emoji from environment + Claude's project detection + current topic -->

Terminal Title Management

Overview

This skill updates the terminal window title to show current context:

  • Work/fun emoji (from $TERMINAL_TITLE_EMOJI environment variable)
  • Project name (intelligently detected by Claude)
  • Current topic (what you're working on)

Format: 💼 ProjectName - Topic

CRITICAL: When to Invoke

MANDATORY invocations:

  1. Session start: Automatically triggered by SessionStart hook. Claude must respond by setting title to "Claude Code" as default topic.
  2. Topic changes: Claude MUST detect and invoke when user shifts to new topic.

Topic change patterns (Claude must detect these):

  • ✅ "let's talk about X" / "can you tell me about Y" → invoke immediately
  • ✅ User switches domains: debugging → documentation, frontend → backend, feature → tests
  • ✅ User starts working on different module/component after sustained discussion
  • ✅ User asks about completely unrelated topic after 3+ exchanges on current topic
  • ❌ Follow-up questions on same topic ("add a comment to that function") → do NOT invoke
  • ❌ Small refinements to current work ("make it blue") → do NOT invoke
  • ❌ Clarifications about current task → do NOT invoke

Claude's responsibility: Actively monitor conversation flow and invoke this skill whenever topic materially shifts. Do not wait for explicit permission.

Example titles:

  • 💼 Skills Repository - Terminal Title
  • 🎉 dotfiles - zsh config
  • 💼 OneOnOne - Firebase Config

Setup Requirements

Required Permission:

This skill runs scripts to update the terminal title. To prevent Claude from prompting for permission every time, add this to your ~/.claude/settings.json:

For Unix/Linux/macOS:

{
  "permissions": {
    "allow": [
      "Bash(bash *skills/scripts/set_title.sh:*)"
    ]
  }
}

For Windows:

{
  "permissions": {
    "allow": [
      "Bash(pwsh *skills/scripts/set_title.ps1:*)"
    ]
  }
}

Why this permission is needed: The skill executes a script that sends terminal escape sequences silently in the background to update your window title without interrupting your workflow.

Cross-platform support: The plugin includes both bash scripts (.sh) for Unix-like systems and PowerShell scripts (.ps1) for Windows. The SessionStart hook automatically detects your OS and uses the appropriate script.

Project Detection Strategy

Primary method: Claude's intelligence

Claude analyzes the current context to determine the project name:

  • Current working directory and recent conversation
  • Files read during the session
  • Git repository information
  • Package metadata (package.json, README.md, etc.)
  • Overall understanding of what the project represents

Key principle: Use intelligent understanding, not just mechanical file parsing.

Detection examples:

  • /Users/dylanr/work/2389/skills → "Skills Repository" (not just "skills")
  • /Users/dylanr/work/2389/oneonone/hosting → "OneOnOne"
  • /Users/dylanr/dotfiles → "dotfiles"
  • /Users/dylanr/projects/my-app → "My App" (humanized from directory)

Supporting evidence Claude checks:

  1. Git remote URL or repository directory name
  2. package.json name field (Node.js projects)
  3. First heading in README.md
  4. Directory basename (last resort)

Fallback: If Claude cannot determine context, use current directory name.

Always include project name - Claude can always determine something meaningful.

Title Formatting

Standard format:

$emoji ProjectName - Topic

Component details:

1. Emoji (from environment):

  • Read from $TERMINAL_TITLE_EMOJI environment variable
  • Set by zsh theme based on directory context
  • Common values: 💼 (work), 🎉 (fun/personal)
  • Fallback: Use 🎉 if variable not set

2. Project Name (from Claude's detection):

  • Human-friendly name, not slug format
  • Proper capitalization (e.g., "OneOnOne" not "oneonone")
  • Always present (use directory name as minimum)

3. Topic (from invocation context):

  • Short description of current work (2-4 words)
  • Provided when skill is invoked
  • Examples: "Terminal Title", "Firebase Config", "CSS Refactor"

Script usage:

bash scripts/set_title.sh "ProjectName" "Topic"

Note: The script automatically reads emoji from environment and handles terminal escape sequences.

Edge cases:

  • No $TERMINAL_TITLE_EMOJI: Use 🎉 as default
  • Very long project/topic: Let terminal handle truncation naturally
  • Empty topic: Use "Claude Code" as default

Workflow

When Claude detects a topic change (or session start), immediately invoke this workflow:

Follow these steps to update the terminal title:

Step 1: Determine project name

Use Claude's understanding of the current codebase and context to determine the project name. Check:

  • What you know about this project from conversation and files
  • Git repository name/remote URL
  • package.json name field (if Node.js project)
  • README.md first heading
  • Directory basename (minimum fallback)

Generate a human-friendly name (e.g., "Skills Repository" not "skills").

Step 2: Get topic from context

The topic should come from:

  • Invocation parameter (if provided explicitly)
  • Current conversation context (what user is working on)
  • Default to "Claude Code" for session start

Step 3: Detect operating system

Before executing the script, detect the operating system:

# Check if running on Windows (check for common Windows indicators)
if [[ "$OSTYPE" == "msys" || "$OSTYPE" == "win32" || "$OSTYPE" == "cygwin" ]]; then
    OS="windows"
else
    OS="unix"
fi

Step 4: Execute title update script (cross-platform)

Find the skill base directory (available in the skill loading message) and execute the appropriate script based on OS:

For Unix/Linux/macOS:

bash <skill-base-dir>/scripts/set_title.sh "Project Name" "Topic"

For Windows:

pwsh -NoProfile -ExecutionPolicy Bypass -File <skill-base-dir>/scripts/set_title.ps1 "Project Name" "Topic"

The scripts handle:

  • Reading emoji from $TERMINAL_TITLE_EMOJI (Unix) or $env:TERMINAL_TITLE_EMOJI (Windows)
  • Sanitizing inputs
  • Formatting and sending terminal escape sequences

Execute this command silently using the Bash tool (pre-approved via permissions).

Step 5: Done

Title updated silently. Continue responding to user without mentioning the title update unless they ask about it.

Examples

Example 1: Session Start in Skills Repository

Context: User starts Claude Code in /Users/dylanr/work/2389/skills

Environment: TERMINAL_TITLE_EMOJI=💼

Execution:

  1. Project: "Skills Repository" (Claude's knowledge)
  2. Topic: "Claude Code" (session start default)
  3. Command: bash /Users/dylanr/work/2389/claude-plugins/terminal-title/skills/scripts/set_title.sh "Skills Repository" "Claude Code"

Result: Terminal title shows 💼 Skills Repository - Claude Code (script reads emoji from environment)


Example 2: Topic Change Detection

Context: Conversation shifts from "terminal title implementation" to "rare VWs"

Claude's detection: User asked "can you tell me about rare vws?" - this is a clear topic shift

Environment: TERMINAL_TITLE_EMOJI=🎉

Claude's action:

  1. Detect topic change (from terminal implementation to VWs)
  2. Invoke terminal-title skill workflow
  3. Project: "Home" (current directory basename)
  4. Topic: "Rare VWs" (from new conversation direction)
  5. Command: bash <skill-base-dir>/scripts/set_title.sh "Home" "Rare VWs"
  6. Continue answering user's question about rare VWs

Result: Terminal title silently updates to 🎉 Home - Rare VWs (script reads emoji from environment)


Example 3: Personal Project Without Environment Variable

Context: User working in ~/projects/dotfiles directory

Environment: TERMINAL_TITLE_EMOJI not set

Execution:

  1. Project: "dotfiles" (from directory)
  2. Topic: "zsh config" (from conversation)
  3. Command: bash <skill-base-dir>/scripts/set_title.sh "dotfiles" "zsh config"

Result: Terminal title shows 🎉 dotfiles - zsh config (script uses 🎉 as fallback when TERMINAL_TITLE_EMOJI not set)


Example 4: Different Project Context

Context: User working in /Users/dylanr/work/2389/oneonone/hosting

Environment: TERMINAL_TITLE_EMOJI=💼

Execution:

  1. Project: "OneOnOne" (Claude knows this project name)
  2. Topic: "Firebase Config" (from conversation)
  3. Command: bash <skill-base-dir>/scripts/set_title.sh "OneOnOne" "Firebase Config"

Result: Terminal title shows 💼 OneOnOne - Firebase Config (script reads emoji from environment)


Example 5: Windows User Session Start

Context: Windows user starts Claude Code in C:\Users\Nat\projects\my-app

Environment: $env:TERMINAL_TITLE_EMOJI=💻 (Windows environment variable)

OS Detection: Claude detects Windows via $OSTYPE check

Execution:

  1. Project: "My App" (from directory name, humanized)
  2. Topic: "Claude Code" (session start default)
  3. Command: pwsh -NoProfile -ExecutionPolicy Bypass -File <skill-base-dir>/scripts/set_title.ps1 "My App" "Claude Code"

Result: Terminal title shows 💻 My App - Claude Code (PowerShell script reads emoji from $env:TERMINAL_TITLE_EMOJI)

Note: The SessionStart hook automatically detected Windows and invoked the PowerShell version of the script instead of the bash version.

レビュー

まだレビューはありません。使ってみた感想をお寄せください。

同じリポジトリのスキル

概要と使いどころ

Trigger for: explaining current FAR, DFARS, or agency-supplement text; determining documented acquisition-policy status for an agency and FAR part; comparing codified text, RFO model text, and agency deviations; tracing acquisition rulemaking; finding procurement comment periods; analyzing public comments; refreshing prior policy analysis; or producing an Acquisition Policy Impact Brief. Use published federal sources and supplied policy documents without making legal or procurement-specific applicability determinations. Do not use for clause selection, market research execution, opportunities, pricing, grants, or cooperative agreements.

日本語の概要は準備中です。原文の説明を表示しています。

aiskillstore/marketplace4322026年10月13日 更新

Systematic framework for extracting actionable items from documents and organizing them into prioritized, trackable checklists. Use when converting reports, meeting notes, audits, or any document with embedded action items into structured TODO lists.

日本語の概要は準備中です。原文の説明を表示しています。

aiskillstore/marketplace4322026年10月13日 更新

Browser automation for AI agents via inference.sh. Navigate web pages, interact with elements using @e refs, take screenshots, record video. Capabilities: web scraping, form filling, clicking, typing, drag-drop, file upload, JavaScript execution. Use for: web automation, data extraction, testing, agent browsing, research. Triggers: browser, web automation, scrape, navigate, click, fill form, screenshot, browse web, playwright, headless browser, web agent, surf internet, record video

日本語の概要は準備中です。原文の説明を表示しています。

aiskillstore/marketplace4322026年10月13日 更新

This skill should be used when the user asks to "create an agent", "add an agent", "write a subagent", "agent frontmatter", "when to use description", "agent examples", "agent tools", "agent colors", "autonomous agent", or needs guidance on agent structure, system prompts, triggering conditions, or agent development best practices for Claude Code plugins.

日本語の概要は準備中です。原文の説明を表示しています。

aiskillstore/marketplace4322026年10月13日 更新

Run AI apps via inference.sh CLI - image generation, video creation, LLMs, search, 3D, Twitter automation. Models: FLUX, Veo, Gemini, Grok, Claude, Seedance, OmniHuman, Tavily, Exa, OpenRouter, and many more. Use when running AI apps, generating images/videos, calling LLMs, web search, or automating Twitter. Triggers: inference.sh, infsh, ai model, run ai, serverless ai, ai api, flux, veo, claude api, image generation, video generation, openrouter, tavily, exa search, twitter api, grok

日本語の概要は準備中です。原文の説明を表示しています。

aiskillstore/marketplace4322026年10月13日 更新

Agile ceremonies and sprint management specialist. Use when running standups, retrospectives, sprint planning, or tracking sprint progress.

日本語の概要は準備中です。原文の説明を表示しています。

aiskillstore/marketplace4322026年10月13日 更新

aiskillstore のスキルをすべて見る

このスキルの問題を報告する