Add ui agent (#343)

* add uipath agent

* readme update

mm_agents/uipath/grounder_client.py

@@ -0,0 +1,43 @@
+import httpx
+import mm_agents.uipath.utils as utils
+import os
+
+class GrounderClient(object):
+    def __init__(self):
+        # Proxy for hosting UI-TARS + UiElementPredictor
+        # Could be replaced with a VLLM server and grounder (UI-TARS) specific processing
+        # Or any other grounder 
+        self.url = ""
+
+    async def predict(
+        self, image_base64: str, action_description: str, action: str | None = None
+    ) -> utils.GroundingOutput:
+        request = utils.GroundingRequest(
+            description=action_description,
+            image_base64=image_base64,
+            action_type=action,
+        )
+        api_key = os.getenv("SERVICE_KEY")
+
+        async with httpx.AsyncClient() as client:
+            response = await client.post(
+                self.url,
+                json={
+                    "image_base64": request.image_base64,
+                    "action_description": request.description,
+                    "action": request.action_type,
+                },
+                headers={
+                    "X-API-KEY": api_key
+                },
+                timeout=100.0,
+            )
+
+        if response.status_code != 200:
+            raise ValueError(f"Prediction failed: {response.text}")
+
+        data = response.json()
+        return utils.GroundingOutput(
+            description=data["description"],
+            position=tuple(data["position"]),
+        )