Page capture tool for visual inspection and documentation. Returns screenshot as embedded image or saves to filesystem.
Scanned 6/1/2026
Install via CLI
openskills install tools-only/X-Skills# Tool Specification: mshtools-browser_screenshot
## Overview
Page capture tool for visual inspection and documentation. Returns screenshot as embedded image or saves to filesystem.
## JSON Schema
```json
{
"type": "object",
"properties": {
"url": {
"type": "string",
"description": "URL to capture (optional, uses current page if omitted)"
},
"citation_id": {
"type": "integer",
"description": "Citation ID of tab to capture (optional)"
},
"download_screenshot_path": {
"type": "string",
"description": "Absolute path to save screenshot (optional, empty string = don't save)"
}
}
}
```
## Streaming Mechanism
- **Capture Method**: CDP Page.captureScreenshot
- **Format**: PNG (default)
- **Viewport**: Current browser viewport only (not full page)
- **Return**: Image data embedded in response or saved to path
## Integration Architecture
### CDP Page Domain
```javascript
Page.captureScreenshot({
format: 'png',
fromSurface: true
})
```
## Usage Patterns
### Capture Current Page
```
browser_visit(url="https://example.com")
browser_screenshot() # Returns inline image
```
### Save to File
```
browser_screenshot(
citation_id=1,
download_screenshot_path="/mnt/kimi/output/screenshot.png"
)
```
### Visual Verification
```
browser_click(citation_id=1, element_index=3)
browser_screenshot(citation_id=1) # Verify result
```
## Limitations
- **Viewport Only**: Captures visible area, not full scrollable page
- **Dynamic Content**: May catch loading states
## Related Tools
- `screenshot_web_full_page`: Capture full scrollable page
No comments yet. Be the first to comment!