Image Operations
All image steps use template PNG files captured with robost-snip.
wait_image
Wait until a template appears on screen.
- wait_image:
template: button.png
timeout_ms: 5000 # default: 10000
threshold: 0.88 # match confidence 0–1 (default: 0.88)
click_image
Wait for a template then click its center.
- click_image:
template: submit.png
timeout_ms: 5000
button: left # left | right | middle (default: left)
offset_x: 0 # pixel offset from center
offset_y: 0
find_image
Search for a template and save its position without clicking.
match_rect
Match a template within a restricted screen region.
wait_no_image
Wait until a template disappears from the screen.
wait_change
Wait until the screen changes (any pixel difference in the region).
- wait_change:
timeout_ms: 5000
region: # optional — watch only this area
x: 0
y: 0
width: 800
height: 600
ocr_match
Wait for text to appear on screen using OCR.
- ocr_match:
contains: "処理完了"
lang: jpn+eng # Tesseract language codes
timeout_ms: 10000
save_as: matched_text
ml_detect
Detect objects using an ONNX model (YOLO-compatible).
- ml_detect:
model: model.onnx
threshold: 0.7
save_as: detections # list of { label, x, y, width, height, confidence }
screenshot_save
Save the current screen to a PNG file.
get_pixel_color
Read the color of a single pixel.
wait_color
Wait until a pixel reaches the specified color.
- wait_color:
x: 500
y: 300
color: "#00FF00"
tolerance: 10 # per-channel tolerance (default: 0)
timeout_ms: 10000
window_control
Bring a window to focus, minimize, maximize, or close it.