Prefer Agent.execute() for interactions that require finding elements visually. Use Computer methods when you know exact coordinates or need precise keyboard input.
click_at()
computer.click_at(x: int, y: int, button: str = "left") -> None
Click at specific screen coordinates.
| Parameter | Type | Default | Description |
|---|
x | int | — | X coordinate |
y | int | — | Y coordinate |
button | str | "left" | "left" or "right" |
def run(params: Params) -> Result:
computer = Computer()
computer.click_at(100, 200) # Left click
computer.click_at(100, 200, button="right") # Right click
move()
computer.move(x: int, y: int) -> None
Move the cursor without clicking.
def run(params: Params) -> Result:
computer = Computer()
computer.move(500, 300)
computer.scroll(direction: str = "down", amount: int = 3, x: int | None = None, y: int | None = None) -> None
Scroll the page.
| Parameter | Type | Default | Description |
|---|
direction | str | "down" | "up" or "down" |
amount | int | 3 | Number of scroll ticks |
x | int | None | None | Scroll at X coordinate |
y | int | None | None | Scroll at Y coordinate |
def run(params: Params) -> Result:
computer = Computer()
computer.scroll() # Scroll down 3 ticks
computer.scroll("up", 5) # Scroll up 5 ticks
computer.scroll("down", 3, x=400, y=300) # Scroll at specific coordinates