GitHub - jaehongpark-agent/claude-code-spinner-verbs

Ever wondered what those quirky loading messages in Claude Code are?

This tool extracts all spinner verbs from the Claude Code CLI binary and saves them as versioned markdown files for history tracking.

Korean (한국어)

How It Works

1. Locates the Claude Code binary on your system
2. Runs strings to dump printable text from the binary
3. Uses seed-based bootstrapping to find the spinner verb array — known rare verbs (e.g. Flibbertigibbeting, Razzmatazzing) pinpoint the right location
4. Parses out all gerund-form words and saves them to words/<version>.md
5. Shows a diff against the previous extraction, so you can spot additions and removals across versions

Usage

# Extract spinner verbs only
python3 extract_spinner_verbs.py

# Extract + generate llms.txt (for AI agents)
python3 build.py

Output example:

Binary:  /Users/you/.local/share/claude/versions/2.1.63
Version: 2.1.63
Seeds:   187 words (12 built-in)

Extracted 187 spinner verbs:

    1. Accomplishing
    2. Actioning
    3. Actualizing
  ...
  187. Zigzagging

Saved to words/2.1.63.md
Saved to llms.txt

Requirements

• Python 3.10+
• Claude Code CLI installed
• strings command available (pre-installed on macOS/Linux)

Project Structure

.
├── extract_spinner_verbs.py   # Extractor (standalone)
├── build.py                   # Build pipeline (extract + llms.txt)
├── llms.txt                   # Auto-generated context for AI agents
└── words/
    ├── 2.1.63.md              # English words by version
    └── 2.1.63_kr.md           # Korean translations by version

Customizing Your Spinner

Want to use your own spinner verbs? See the Customization Guide to learn how to configure spinnerVerbs in settings.json.

Sample Words

Some highlights from the 187 spinner verbs found in v2.1.63:

License

MIT