topai_agentvision_crop
Computes optimal Region-of-Interest bounding boxes for UI buttons, tables, charts, or text blocks, cutting vision token consumption by up to 75% before images are sent to vision-capable LLMs.
Instructions
Computes optimal Region-of-Interest (ROI) bounding boxes for UI buttons, tables, charts, or text blocks in an image to reduce multimodal vision token consumption by up to 75%. Use when preparing screenshots or document images for vision-capable LLMs so only the relevant region is sent. Do not use for taking the initial webpage screenshot (use topai_headless_screenshot first). Stateless, read-only, idempotent visual ROI analysis.
Tool call template
Replace each placeholder with a value matching the input schema. Only required parameters are included; add optional parameters as needed.
{
"id": 1,
"jsonrpc": "2.0",
"method": "tools/call",
"params": {
"arguments": {
"image_url": "<image_url>"
},
"name": "topai_agentvision_crop"
}
}Input Schema
| Name | Required | Description | Default |
|---|---|---|---|
| image_url | Yes | Public HTTP/HTTPS URL of the source image to analyze for Region-of-Interest cropping. | |
| target_element | No | Specific visual element category to isolate: 'auto', 'button', 'table', 'text_block', 'chart', or 'header' (default: 'auto'). |