38 lines
1.1 KiB
Markdown
38 lines
1.1 KiB
Markdown
---
|
|
name: ollama-vision
|
|
description: Automated Vision pipeline using Peekaboo for capture and Ollama (llama3.2-vision) for analysis.
|
|
homepage: https://ollama.com
|
|
metadata: { "openclaw": { "emoji": "👁️", "requires": { "bins": ["peekaboo", "ollama"] } } }
|
|
---
|
|
|
|
# Ollama Vision
|
|
|
|
Bridge macOS screen capture directly to local vision models.
|
|
|
|
## Usage
|
|
|
|
```bash
|
|
# General screen description
|
|
./scripts/vision-bridge.sh "What's on my screen?"
|
|
|
|
# Focus on a specific app
|
|
./scripts/vision-bridge.sh "Describe this window" "Slack"
|
|
|
|
# Higher detail (retina)
|
|
./scripts/vision-bridge.sh "Read the text in this terminal" "iTerm2" true
|
|
```
|
|
|
|
## Tools
|
|
|
|
### vision.see
|
|
Capture the screen or a specific app and describe what is visible using Ollama Llama 3.2 Vision.
|
|
|
|
**Run configuration:**
|
|
- `command`: `/Users/MWsatwareAG/internal/mw-macbook-pro/scripts/vision-bridge.sh`
|
|
- `args`: ["{{prompt}}", "{{app}}", "{{retina}}"]
|
|
|
|
**Arguments:**
|
|
- `prompt`: (string) What to look for or ask about the image.
|
|
- `app`: (optional string) Name or bundle ID of the app to focus on.
|
|
- `retina`: (optional boolean, default true) Capture at high resolution.
|