feat(vom): canvas position calculation and projection

crates/bsk-protocol/schema/tool_click_params.json

@@ -16,6 +16,13 @@
         }
       ]
     },
+    "capture_id": {
+      "description": "Short-lived capture id returned by a visual Surface screenshot.",
+      "type": [
+        "string",
+        "null"
+      ]
+    },
     "click_count": {
       "description": "Number of consecutive mouse presses (double-click = 2).",
       "type": [
@@ -25,6 +32,22 @@
       "format": "uint32",
       "minimum": 1.0
     },
+    "image_x": {
+      "description": "X coordinate in the capture image's pixel coordinate space.",
+      "type": [
+        "number",
+        "null"
+      ],
+      "format": "double"
+    },
+    "image_y": {
+      "description": "Y coordinate in the capture image's pixel coordinate space.",
+      "type": [
+        "number",
+        "null"
+      ],
+      "format": "double"
+    },
     "modifiers": {
       "type": [
         "array",

crates/bsk-protocol/schema/tool_click_result.json

@@ -8,6 +8,12 @@
     "y"
   ],
   "properties": {
+    "capture_id": {
+      "type": [
+        "string",
+        "null"
+      ]
+    },
     "dialogs": {
       "description": "Native JS dialogs observed and auto-handled during this call.",
       "type": "array",
@@ -15,6 +21,20 @@
         "$ref": "#/definitions/JavaScriptDialogInfo"
       }
     },
+    "image_x": {
+      "type": [
+        "number",
+        "null"
+      ],
+      "format": "double"
+    },
+    "image_y": {
+      "type": [
+        "number",
+        "null"
+      ],
+      "format": "double"
+    },
     "tab_id": {
       "type": "integer",
       "format": "int64"

crates/bsk-protocol/schema/tool_screenshot_result.json

@@ -8,6 +8,16 @@
     "tab_id"
   ],
   "properties": {
+    "capture": {
+      "anyOf": [
+        {
+          "$ref": "#/definitions/SurfaceCaptureInfo"
+        },
+        {
+          "type": "null"
+        }
+      ]
+    },
     "dialogs": {
       "type": "array",
       "items": {
@@ -109,6 +119,31 @@
         "prompt",
         "beforeunload"
       ]
+    },
+    "SurfaceCaptureInfo": {
+      "type": "object",
+      "required": [
+        "coordinate_space",
+        "expires_at",
+        "id",
+        "surface_ref"
+      ],
+      "properties": {
+        "coordinate_space": {
+          "type": "string"
+        },
+        "expires_at": {
+          "type": "integer",
+          "format": "uint64",
+          "minimum": 0.0
+        },
+        "id": {
+          "type": "string"
+        },
+        "surface_ref": {
+          "type": "string"
+        }
+      }
     }
   }
 }

crates/bsk-protocol/src/tools/interaction.rs

@@ -67,6 +67,15 @@ pub struct ClickParams {
     #[serde(default, skip_serializing_if = "Option::is_none")]
     #[schemars(range(min = 1))]
     pub timeout_ms: Option<u32>,
+    /// Short-lived capture id returned by a visual Surface screenshot.
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub capture_id: Option<String>,
+    /// X coordinate in the capture image's pixel coordinate space.
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub image_x: Option<f64>,
+    /// Y coordinate in the capture image's pixel coordinate space.
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub image_y: Option<f64>,
 }
 
 #[derive(Debug, Clone, PartialEq, Serialize, Deserialize, JsonSchema)]
@@ -80,6 +89,12 @@ pub struct ClickResult {
     /// agents can correlate with a follow-up `tool.screenshot`.
     pub x: f64,
     pub y: f64,
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub capture_id: Option<String>,
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub image_x: Option<f64>,
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub image_y: Option<f64>,
     /// Native JS dialogs observed and auto-handled during this call.
     #[serde(default, skip_serializing_if = "Vec::is_empty")]
     pub dialogs: Vec<JavaScriptDialogInfo>,
@@ -288,6 +303,9 @@ mod tests {
             click_count: Some(1),
             modifiers: Some(vec![KeyModifier::Ctrl]),
             timeout_ms: Some(5_000),
+            capture_id: Some("sc_test".into()),
+            image_x: Some(160.0),
+            image_y: Some(90.0),
         };
         let v = serde_json::to_value(&p).unwrap();
         assert_eq!(v.get("ref").and_then(|v| v.as_str()), Some("@e3"));

crates/bsk-protocol/src/tools/observation.rs

@@ -208,10 +208,20 @@ pub struct ScreenshotResult {
     /// Always `"png"` in v0.1; reserved for future JPEG / WebP support.
     pub format: String,
     pub tab_id: i64,
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub capture: Option<SurfaceCaptureInfo>,
     #[serde(default, skip_serializing_if = "Vec::is_empty")]
     pub dialogs: Vec<JavaScriptDialogInfo>,
 }
 
+#[derive(Debug, Clone, PartialEq, Serialize, Deserialize, JsonSchema)]
+pub struct SurfaceCaptureInfo {
+    pub id: String,
+    pub surface_ref: String,
+    pub coordinate_space: String,
+    pub expires_at: u64,
+}
+
 #[cfg(test)]
 mod tests {
     use super::*;
@@ -293,6 +303,12 @@ mod tests {
             height: 600,
             format: "png".into(),
             tab_id: 7,
+            capture: Some(SurfaceCaptureInfo {
+                id: "sc_test".into(),
+                surface_ref: "@e3".into(),
+                coordinate_space: "capture-image-pixel".into(),
+                expires_at: 1_700_000_030_000,
+            }),
             dialogs: vec![],
         };
         let v = serde_json::to_value(&r).unwrap();