Files
sci-gui-agent-benchmark/evaluation_examples/examples/os/ec4e3f68-9ea4-4c18-a5c9-69f89d1178b3.json
yuanmengqi e433f35c1f feat: standardize configuration fields across all evaluation examples
- Add `fixed_ip` field to all 369 JSON files in examples directory
  - Set to `true` for 8 files listed in google_chrome.json multi_apps
  - Set to `false` for remaining 361 files
- Add `possibility_of_env_change` field to 363 JSON files missing this field
  - Set to "low" for newly added fields
  - Preserve existing values (4 medium, 2 high) for 6 files that already had this field

This ensures consistent configuration schema across all evaluation examples
while maintaining backward compatibility with existing settings.
2025-07-16 13:45:34 +00:00

62 lines
1.6 KiB
JSON

{
"id": "ec4e3f68-9ea4-4c18-a5c9-69f89d1178b3",
"snapshot": "os",
"instruction": "Can you remove vim from favorite app in 'favorites'?",
"source": "https://www.youtube.com/watch?v=D4WyNjt_hbQ&t=2s",
"config": [
{
"type": "execute",
"parameters": {
"command": "export DBUS_SESSION_BUS_ADDRESS='unix:path=/run/user/1000/bus'\ngsettings set org.gnome.shell favorite-apps \"['thunderbird.desktop']\"",
"shell": true
}
},
{
"type": "sleep",
"parameters": {
"seconds": 0.5
}
},
{
"type": "execute",
"parameters": {
"command": "export DBUS_SESSION_BUS_ADDRESS='unix:path=/run/user/1000/bus'\ngsettings set org.gnome.shell favorite-apps \"['thunderbird.desktop', 'vim.desktop', 'google-chrome.desktop']\"",
"shell": true
}
},
{
"type": "execute",
"parameters": {
"command": [
"python",
"-c",
"import pyautogui; import time; pyautogui.click({SCREEN_WIDTH_HALF}, {SCREEN_HEIGHT_HALF}); time.sleep(0.5);"
]
}
}
],
"trajectory": "trajectories/",
"related_apps": [
"os"
],
"evaluator": {
"func": "check_gnome_favorite_apps",
"result": {
"type": "vm_command_line",
"command": "export DBUS_SESSION_BUS_ADDRESS='unix:path=/run/user/1000/bus'\ngsettings get org.gnome.shell favorite-apps",
"shell": true
},
"expected": {
"type": "rule",
"rules": {
"expected": [
"google-chrome.desktop",
"thunderbird.desktop"
]
}
}
},
"proxy": false,
"fixed_ip": false,
"possibility_of_env_change": "low"
}