Pi extension that intercepts image reads when using glm-4.7 and sends them to glm-4.6v for detailed analysis
npm install pi-glm-image-summaryA pi extension that intercepts image reads when using non-vision GLM models and sends them to glm-4.6v for detailed analysis.
GLM text models (glm-4.6, glm-4.7, glm-4.7-flash) have no vision capabilities. GLM-4.6v does. This extension automatically detects when you're using a non-vision GLM model and intercepts image reads, sending them to glm-4.6v for comprehensive analysis.
- Automatic image interception: When using glm-4.7/glm-4.7-long, image file reads are automatically redirected to glm-4.6v for analysis
- Comprehensive analysis: Extracts text content, visual elements, technical details, and more
- Manual analysis command: /analyze-image to manually analyze any image
Install globally:
``bash`
pi install npm:pi-glm-image-summary
Or install for a specific project (writes to .pi/settings.json):
`bash`
pi install -l npm:pi-glm-image-summary
To try it without installing:
`bash`
pi -e npm:pi-glm-image-summary
Once installed, the extension loads automatically when you start pi:
`bash`
pi --provider zai --model glm-4.7
When the extension detects:
1. Current model is glm-4.7 or glm-4.7-long
2. A file being read is an image (jpg, jpeg, png, gif, webp)
It will automatically spawn a subprocess with glm-4.6v to analyze the image and return a detailed summary.
Use the /analyze-image command to analyze any image:
```
/analyze-image ./screenshot.png
- JPEG (.jpg, .jpeg)
- PNG (.png)
- GIF (.gif)
- WebP (.webp)
The extension uses the ZAI provider for the vision model. Make sure you have proper API credentials configured.