| name | claude-image-analyzer |
| description | Describe images in detail using Claude Code CLI. Use when you need a detailed description of an image's content, especially when the standard image analysis tools are unavailable. Works with local image files (jpg, png, etc.) by spawning Claude Code with the image path. |
Claude Image Analyzer
Describes images in detail using the Claude Code CLI as a fallback when standard image analysis APIs are unavailable.
When to Use
- When OpenAI/Gemini image APIs fail due to quota or auth issues
- When you need a detailed description of an image
- For complex images that benefit from Claude's reasoning
- As a fallback when standard
image tool fails
How It Works
- Takes a local image file path as input
- Spawns Claude Code CLI with the image
- Claude Code reads and describes the image in detail
- Returns the text description
Usage
Basic Usage
scripts/claude-image-analyzer.sh <image-path>
From Agent Session
When an image is downloaded (e.g., from Discord attachments):
IMAGE_PATH="<media-inbound-dir>/image.jpg"
scripts/claude-image-analyzer.sh "$IMAGE_PATH"
Important Notes
- Requires Claude Code CLI (
claude) to be installed and on PATH
- Interactive: Claude Code may prompt for confirmation when accessing new directories
- Slower than API: Spawns a full Claude Code session, takes 10-30 seconds
- Best-effort: If expect is not installed, falls back to manual interaction mode
Alternative
If this skill is unavailable or too slow, consider:
- Using the standard
image tool with different models
- Switching image model in config to Gemini or other providers
