update documentation; emphasize the importance of using grid for estimating coordinates

Author: dcloud347

configs/examples/general_purpose.yaml

@@ -92,9 +92,9 @@ web_agent:
     - Base your judgement on visual appearance and behaviour, not on source code.
     - If the page shows a blank white page or an error message, check console log and report that clearly.
     
-    ## Estimating Coordinates with the Grid
+    ## Estimating Coordinates with the Grid (CRITICAL)
     When you need to click, type, or interact with a specific element and you are unsure of its exact pixel position:
-    1. Take a screenshot with `grid=true` (optionally set `grid_spacing` for finer resolution, e.g. 100).
+    1. ALWAYS take a screenshot with `grid=true` (optionally set `grid_spacing` for finer resolution, e.g. 100).
     2. The screenshot will show red grid lines with x/y labels along the edges.
     3. Read the grid labels to estimate the x/y pixel coordinates of the target element.
     4. Use those coordinates to estimate the position in your click/type/hover call.

configs/examples/swebench_multimodal.yaml

@@ -114,9 +114,9 @@ web_agent:
     - Base your judgement on visual appearance and behaviour, not on source code.
     - If the page shows a blank white page or an error message, check console log and report that clearly.
     
-    ## Estimating Coordinates with the Grid
+    ## Estimating Coordinates with the Grid (CRITICAL)
     When you need to click, type, or interact with a specific element and you are unsure of its exact pixel position:
-    1. Take a screenshot with `grid=true` (optionally set `grid_spacing` for finer resolution, e.g. 100).
+    1. ALWAYS take a screenshot with `grid=true` (optionally set `grid_spacing` for finer resolution, e.g. 100).
     2. The screenshot will show red grid lines with x/y labels along the edges.
     3. Read the grid labels to estimate the x/y pixel coordinates of the target element.
     4. Use those coordinates to estimate the position in your click/type/hover call.

configs/examples/webgen_bench.yaml

@@ -90,12 +90,12 @@ web_agent:
     - Base your judgement on visual appearance and behaviour, not on source code.
     - If the page shows a blank white page or an error message, check console logs and report clearly.
 
-    ## Estimating Coordinates with the Grid
+    ## Estimating Coordinates with the Grid (CRITICAL)
     When you need to click, type, or interact with a specific element and you are unsure of its exact pixel position:
-    1. Take a screenshot with `grid=true` (optionally set `grid_spacing` for finer resolution, e.g. 100).
+    1. ALWAYS take a screenshot with `grid=true` (optionally set `grid_spacing` for finer resolution, e.g. 100).
     2. The screenshot will show red grid lines with x/y labels along the edges.
     3. Read the grid labels to estimate the x/y pixel coordinates of the target element.
-    4. Use those coordinates in your click/type call.
+    4. Use those coordinates to estimate the position in your click/type/hover call.
     You do NOT need to use the grid for every screenshot — only when you need to pinpoint coordinates.
 
     ## Actively interact with the page