universal-mcp-applications 0.1.33__py3-none-any.whl → 0.1.39rc8__py3-none-any.whl

universal_mcp/applications/browser_use/README.md

@@ -1 +1,15 @@
-# Browser Use
+# BrowserUseApp MCP Server
+
+An MCP Server for the BrowserUseApp API.
+
+## 🛠️ Tool List
+
+This is automatically generated from OpenAPI schema for the BrowserUseApp API.
+
+
+| Tool | Description |
+|------|-------------|
+| `browser_task` | Creates and starts a browser automation task in the background. Unlike synchronous tools, this returns immediately with a Task ID so the agent does not block while the browser works. You must use `get_browser_task_status` to check progress and retrieve the final result. |
+| `get_browser_task_status` | Checks the status of a browser automation task. Use this to poll for completion and retrieve the final output of a background task. |
+| `list_tasks` | Lists recent browser automation tasks. Useful for finding lost task IDs or auditing recent activity. |
+| `stop_task` | Stops a running browser automation task. Use this to cancel a stuck or long-running task. |

universal_mcp/applications/browser_use/__init__.py

@@ -0,0 +1 @@
+from .app import BrowserUseApp

universal_mcp/applications/browser_use/app.py

@@ -1,9 +1,9 @@
-from typing import Annotated
+from typing import Annotated, Literal
 
 from universal_mcp.applications.application import APIApplication
 from universal_mcp.integrations import Integration
 
-from browser_use_sdk import BrowserUse
+from browser_use_sdk import AsyncBrowserUse
 
 
 class BrowserUseApp(APIApplication):
@@ -12,60 +12,122 @@ class BrowserUseApp(APIApplication):
         self._browser_client = None
 
     @property
-    def browser_client(self) -> BrowserUse:
+    def browser_client(self) -> AsyncBrowserUse:
         if self._browser_client is not None:
             return self._browser_client
         if not self.integration:
             raise ValueError("Integration is required but not provided")
         credentials = self.integration.get_credentials()
-        api_key = (
-            credentials.get("api_key")
-            or credentials.get("API_KEY")
-            or credentials.get("apiKey")
-        )
+        api_key = credentials.get("api_key") or credentials.get("API_KEY") or credentials.get("apiKey")
         if not api_key:
             raise ValueError("API key not found in integration credentials")
-        self._browser_client = BrowserUse(api_key=api_key)
+        self._browser_client = AsyncBrowserUse(api_key=api_key)
         return self._browser_client
 
     async def browser_task(
         self,
-        task: Annotated[str, "What you want the browser to do"],
-        max_steps: Annotated[int, "Max actions to take (1-10, default: 8)"] = 8,
-        llm: Annotated[str, "The LLM to use for the task"] = "gpt-4.1",
+        task: Annotated[str, "The objective of the task"],
+        max_steps: Annotated[int, "Max actions to take (default: 8)"] = 8,
+        llm: Annotated[
+            Literal[
+                "gpt-4.1",
+                "gpt-4.1-mini",
+                "o4-mini",
+                "o3",
+                "gemini-2.5-flash",
+                "gemini-2.5-pro",
+                "claude-sonnet-4-20250514",
+                "gpt-4o",
+                "gpt-4o-mini",
+                "llama-4-maverick-17b-128e-instruct",
+                "claude-3-7-sonnet-20250219",
+            ],
+            "The LLM to use",
+        ] = "gpt-4o",
+        session_id: Annotated[str | None, "Session ID to continue an existing session"] = None,
     ) -> dict:
         """
-        Creates and runs a browser automation task.
+        Creates and starts a browser automation task in the background. Unlike synchronous tools, this returns immediately with a Task ID so the agent does not block while the browser works. You must use `get_browser_task_status` to check progress and retrieve the final result.
 
         Args:
             task (str): The detailed description of the task for the browser to perform.
             max_steps (int, optional): The maximum number of actions the browser can take. Defaults to 8.
-            llm (str, optional): The language model to use for interpreting the task. Defaults to "gpt-4.1".
+            llm (str, optional): The language model to use. Defaults to "gpt-4o".
+            session_id (str, optional): The ID of an existing session to use.
 
         Returns:
-            dict: The result of the completed task, including output and other metadata.
+            dict: The task creation details including the task ID.
+        
+        Tags:
+            browser, automation, background, task, web, research, create, important
         """
-        created_task = self.browser_client.tasks.create_task(
-            llm=llm, task=task, max_steps=max_steps
+        created_task = await self.browser_client.tasks.create_task(
+            llm=llm, task=task, max_steps=max_steps, session_id=session_id
         )
-        result = created_task.complete()
-        return result.model_dump()
+
+        return created_task.model_dump()
 
     async def get_browser_task_status(
         self,
         task_id: Annotated[str, "Task ID to check"],
     ) -> dict:
         """
-        Checks task progress with smart polling.
-
+        Checks the status of a browser automation task. Use this to poll for completion and retrieve the final output of a background task.
+        
         Args:
             task_id (str): The ID of the task to check.
 
         Returns:
-            dict: The current status and details of the task.
+            dict: The current task details including status ('created', 'running', 'completed', 'stopped', 'failed') and output (if completed).
+
+        Tags:
+            status, poll, browser, task, monitoring, output, important
         """
-        task = self.browser_client.tasks.get_task(task_id)
+        task = await self.browser_client.tasks.get_task(task_id)
         return task.model_dump()
 
+    async def list_tasks(
+        self,
+        page_size: Annotated[int, "Number of tasks to return (default: 10)"] = 10,
+    ) -> dict:
+        """
+        Lists recent browser automation tasks. Useful for finding lost task IDs or auditing recent activity.
+
+        Args:
+            page_size (int, optional): The number of tasks to retrieve. Defaults to 10.
+
+        Returns:
+            dict: A list of tasks with their basic details.
+
+        Tags:
+            list, history, browser, tasks, audit
+        """
+        tasks = await self.browser_client.tasks.list_tasks(page_size=page_size)
+        return tasks.model_dump()
+
+    async def stop_task(
+        self,
+        task_id: Annotated[str, "Task ID to stop"],
+    ) -> dict:
+        """
+        Stops a running browser automation task. Use this to cancel a stuck or long-running task.
+
+        Args:
+            task_id (str): The ID of the task to stop.
+
+        Returns:
+            dict: The updated task details showing the stopped status.
+
+        Tags:
+            stop, cancel, browser, task, control
+        """
+        updated_task = await self.browser_client.tasks.update_task(task_id, action="stop")
+        return updated_task.model_dump()
+
     def list_tools(self):
-        return [self.browser_task, self.get_browser_task_status]
+        return [
+            self.browser_task,
+            self.get_browser_task_status,
+            self.list_tasks,
+            self.stop_task,
+        ]