anchorbrowser 0.3.2__tar.gz → 0.3.4__tar.gz

anchorbrowser-0.3.4/.release-please-manifest.json

@@ -0,0 +1,3 @@
+{
+  ".": "0.3.4"
+}

{anchorbrowser-0.3.2 → anchorbrowser-0.3.4}/CHANGELOG.md

@@ -1,5 +1,27 @@
 # Changelog
 
+## 0.3.4 (2025-10-31)
+
+Full Changelog: [v0.3.3...v0.3.4](https://github.com/anchorbrowser/AnchorBrowser-SDK-Python/compare/v0.3.3...v0.3.4)
+
+## 0.3.3 (2025-10-31)
+
+Full Changelog: [v0.3.2...v0.3.3](https://github.com/anchorbrowser/AnchorBrowser-SDK-Python/compare/v0.3.2...v0.3.3)
+
+### Features
+
+* Enhanced agent features documentation ([6ebf420](https://github.com/anchorbrowser/AnchorBrowser-SDK-Python/commit/6ebf4206b4e60185adc696d63c8847dcebd45c83))
+
+
+### Bug Fixes
+
+* **client:** close streams without requiring full consumption ([3ff9aa7](https://github.com/anchorbrowser/AnchorBrowser-SDK-Python/commit/3ff9aa743c874d2ecdbf3884fc630d1c81abc32c))
+
+
+### Chores
+
+* **internal/tests:** avoid race condition with implicit client cleanup ([5883b7c](https://github.com/anchorbrowser/AnchorBrowser-SDK-Python/commit/5883b7ce49e1d130b094b44bee109b9ae20c27a9))
+
 ## 0.3.2 (2025-10-27)
 
 Full Changelog: [v0.3.1...v0.3.2](https://github.com/anchorbrowser/AnchorBrowser-SDK-Python/compare/v0.3.1...v0.3.2)

{anchorbrowser-0.3.2 → anchorbrowser-0.3.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: anchorbrowser
-Version: 0.3.2
+Version: 0.3.4
 Summary: The official Python library for the anchorbrowser API
 Project-URL: Homepage, https://github.com/anchorbrowser/AnchorBrowser-SDK-Python
 Project-URL: Repository, https://github.com/anchorbrowser/AnchorBrowser-SDK-Python

{anchorbrowser-0.3.2 → anchorbrowser-0.3.4}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "anchorbrowser"
-version = "0.3.2"
+version = "0.3.4"
 description = "The official Python library for the anchorbrowser API"
 dynamic = ["readme"]
 license = "Apache-2.0"

{anchorbrowser-0.3.2 → anchorbrowser-0.3.4}/src/anchorbrowser/_streaming.py

@@ -57,9 +57,8 @@ class Stream(Generic[_T]):
         for sse in iterator:
             yield process_data(data=sse.json(), cast_to=cast_to, response=response)
 
-        # Ensure the entire stream is consumed
-        for _sse in iterator:
-            ...
+        # As we might not fully consume the response stream, we need to close it explicitly
+        response.close()
 
     def __enter__(self) -> Self:
         return self
@@ -121,9 +120,8 @@ class AsyncStream(Generic[_T]):
         async for sse in iterator:
             yield process_data(data=sse.json(), cast_to=cast_to, response=response)
 
-        # Ensure the entire stream is consumed
-        async for _sse in iterator:
-            ...
+        # As we might not fully consume the response stream, we need to close it explicitly
+        await response.aclose()
 
     async def __aenter__(self) -> Self:
         return self

{anchorbrowser-0.3.2 → anchorbrowser-0.3.4}/src/anchorbrowser/_version.py

@@ -1,4 +1,4 @@
 # File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
 
 __title__ = "anchorbrowser"
-__version__ = "0.3.2"  # x-release-please-version
+__version__ = "0.3.4"  # x-release-please-version

{anchorbrowser-0.3.2 → anchorbrowser-0.3.4}/src/anchorbrowser/resources/agent.py

@@ -47,6 +47,7 @@ class AgentResource(SyncAPIResource):
         *,
         session_options: Optional[Session] = None,
         task_options: Optional[AgentTaskParams] = None,
+        session_id: Optional[str] = None,
     ) -> str:
         """Execute an AI agent task within a browser session.
 
@@ -62,12 +63,19 @@ class AgentResource(SyncAPIResource):
         Returns:
             str: The result of the AI agent task execution.
         """
-        session = self._client.sessions.create(session=session_options or {})
-        if not session.data or not session.data.id:
-            raise ValueError("Failed to create session: No session ID returned")
+        if session_id:
+            retrieved_session = self._client.sessions.retrieve(session_id)
+            if not retrieved_session or not retrieved_session or not retrieved_session.session_id:
+                raise ValueError("Failed to retrieve session: No session ID returned")
+            actual_session_id = retrieved_session.session_id
+        else:
+            created_session = self._client.sessions.create(session=session_options or {})
+            if not created_session or not created_session.data or not created_session.data.id:
+                raise ValueError("Failed to create session: No session ID returned")
+            actual_session_id = created_session.data.id
 
         with BrowserSetup(
-            session_id=session.data.id,
+            session_id=actual_session_id,
             base_url=str(self._client.base_url),
             api_key=self._client.api_key,
         ) as browser_setup:
@@ -228,6 +236,7 @@ class AsyncAgentResource(AsyncAPIResource):
         *,
         session_options: Optional[Session] = None,
         task_options: Optional[AgentTaskParams] = None,
+        session_id: Optional[str] = None,
     ) -> str:
         """Execute an AI agent task within a browser session.
 
@@ -243,12 +252,19 @@ class AsyncAgentResource(AsyncAPIResource):
         Returns:
             str: The result of the AI agent task execution.
         """
-        session = await self._client.sessions.create(session=session_options or {})
-        if not session.data or not session.data.id:
-            raise ValueError("Failed to create session: No session ID returned")
+        if session_id:
+            retrieved_session = await self._client.sessions.retrieve(session_id)
+            if not retrieved_session or not retrieved_session.session_id:
+                raise ValueError("Failed to retrieve session: No session ID returned")
+            actual_session_id = retrieved_session.session_id
+        else:
+            created_session = await self._client.sessions.create(session=session_options or {})
+            if not created_session or not created_session.data or not created_session.data.id:
+                raise ValueError("Failed to create session: No session ID returned")
+            actual_session_id = created_session.data.id
 
         browser_setup = BrowserSetup(
-            session_id=session.data.id,
+            session_id=actual_session_id,
             base_url=str(self._client.base_url),
             api_key=self._client.api_key,
         )

{anchorbrowser-0.3.2 → anchorbrowser-0.3.4}/src/anchorbrowser/resources/sessions/sessions.py

@@ -42,6 +42,7 @@ from .keyboard import (
     AsyncKeyboardResourceWithStreamingResponse,
 )
 from ..._compat import cached_property
+from ..._models import construct_type
 from .clipboard import (
     ClipboardResource,
     AsyncClipboardResource,
@@ -214,13 +215,21 @@ class SessionsResource(SyncAPIResource):
         """
         if not session_id:
             raise ValueError(f"Expected a non-empty value for `session_id` but received {session_id!r}")
-        return self._get(
+        
+        # Get raw response to unwrap the data field
+        raw_response = cast(dict[str, object], self._get(
             f"/v1/sessions/{session_id}",
             options=make_request_options(
                 extra_headers=extra_headers, extra_query=extra_query, extra_body=extra_body, timeout=timeout
             ),
-            cast_to=SessionRetrieveResponse,
-        )
+            cast_to=object,  # Get as raw object (dict) to unwrap
+        ))
+        
+        # Unwrap data field if API returns { data: {...} }
+        data: object = raw_response.get("data", raw_response)
+        
+        # Construct the response type from the unwrapped data
+        return cast(SessionRetrieveResponse, construct_type(type_=SessionRetrieveResponse, value=data))
 
     def delete(
         self,
@@ -751,13 +760,21 @@ class AsyncSessionsResource(AsyncAPIResource):
         """
         if not session_id:
             raise ValueError(f"Expected a non-empty value for `session_id` but received {session_id!r}")
-        return await self._get(
+        
+        # Get raw response to unwrap the data field
+        raw_response = cast(dict[str, object], await self._get(
             f"/v1/sessions/{session_id}",
             options=make_request_options(
                 extra_headers=extra_headers, extra_query=extra_query, extra_body=extra_body, timeout=timeout
             ),
-            cast_to=SessionRetrieveResponse,
-        )
+            cast_to=object,  # Get as raw object (dict) to unwrap
+        ))
+        
+        # Unwrap data field if API returns { data: {...} }
+        data: object = raw_response.get("data", raw_response)
+        
+        # Construct the response type from the unwrapped data
+        return cast(SessionRetrieveResponse, construct_type(type_=SessionRetrieveResponse, value=data))
 
     async def delete(
         self,

{anchorbrowser-0.3.2 → anchorbrowser-0.3.4}/src/anchorbrowser/resources/tools.py

@@ -2,6 +2,7 @@
 
 from __future__ import annotations
 
+from typing import Dict
 from typing_extensions import Literal
 
 import httpx
@@ -128,10 +129,14 @@ class ToolsResource(SyncAPIResource):
         prompt: str,
         session_id: str | Omit = omit,
         agent: Literal["browser-use", "openai-cua"] | Omit = omit,
+        detect_elements: bool | Omit = omit,
         highlight_elements: bool | Omit = omit,
+        human_intervention: bool | Omit = omit,
+        max_steps: int | Omit = omit,
         model: str | Omit = omit,
         output_schema: object | Omit = omit,
         provider: Literal["openai", "gemini", "groq", "azure", "xai"] | Omit = omit,
+        secret_values: Dict[str, str] | Omit = omit,
         url: str | Omit = omit,
         # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs.
         # The extra values given here take precedence over values defined on the client or passed to this method.
@@ -152,8 +157,16 @@ class ToolsResource(SyncAPIResource):
 
           agent: The AI agent to use for task completion. Defaults to browser-use.
 
+          detect_elements: Enable element detection for better interaction accuracy. Improves the agent's
+              ability to identify and interact with UI elements.
+
           highlight_elements: Whether to highlight elements during task execution for better visibility.
 
+          human_intervention: Allow human intervention during task execution. When enabled, the agent can
+              request human input for ambiguous situations.
+
+          max_steps: Maximum number of steps the agent can take to complete the task. Defaults to 25.
+
           model: The specific model to use for task completion. see our
               [models](/agentic-browser-control/ai-task-completion#available-models) page for
               more information.
@@ -162,6 +175,9 @@ class ToolsResource(SyncAPIResource):
 
           provider: The AI provider to use for task completion.
 
+          secret_values: Secret values to pass to the agent for secure credential handling. Keys and
+              values are passed as environment variables to the agent.
+
           url: The URL of the webpage. If not provided, the tool will use the current page in
               the session.
 
@@ -179,10 +195,14 @@ class ToolsResource(SyncAPIResource):
                 {
                     "prompt": prompt,
                     "agent": agent,
+                    "detect_elements": detect_elements,
                     "highlight_elements": highlight_elements,
+                    "human_intervention": human_intervention,
+                    "max_steps": max_steps,
                     "model": model,
                     "output_schema": output_schema,
                     "provider": provider,
+                    "secret_values": secret_values,
                     "url": url,
                 },
                 tool_perform_web_task_params.ToolPerformWebTaskParams,
@@ -381,10 +401,14 @@ class AsyncToolsResource(AsyncAPIResource):
         prompt: str,
         session_id: str | Omit = omit,
         agent: Literal["browser-use", "openai-cua"] | Omit = omit,
+        detect_elements: bool | Omit = omit,
         highlight_elements: bool | Omit = omit,
+        human_intervention: bool | Omit = omit,
+        max_steps: int | Omit = omit,
         model: str | Omit = omit,
         output_schema: object | Omit = omit,
         provider: Literal["openai", "gemini", "groq", "azure", "xai"] | Omit = omit,
+        secret_values: Dict[str, str] | Omit = omit,
         url: str | Omit = omit,
         # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs.
         # The extra values given here take precedence over values defined on the client or passed to this method.
@@ -405,8 +429,16 @@ class AsyncToolsResource(AsyncAPIResource):
 
           agent: The AI agent to use for task completion. Defaults to browser-use.
 
+          detect_elements: Enable element detection for better interaction accuracy. Improves the agent's
+              ability to identify and interact with UI elements.
+
           highlight_elements: Whether to highlight elements during task execution for better visibility.
 
+          human_intervention: Allow human intervention during task execution. When enabled, the agent can
+              request human input for ambiguous situations.
+
+          max_steps: Maximum number of steps the agent can take to complete the task. Defaults to 25.
+
           model: The specific model to use for task completion. see our
               [models](/agentic-browser-control/ai-task-completion#available-models) page for
               more information.
@@ -415,6 +447,9 @@ class AsyncToolsResource(AsyncAPIResource):
 
           provider: The AI provider to use for task completion.
 
+          secret_values: Secret values to pass to the agent for secure credential handling. Keys and
+              values are passed as environment variables to the agent.
+
           url: The URL of the webpage. If not provided, the tool will use the current page in
               the session.
 
@@ -432,10 +467,14 @@ class AsyncToolsResource(AsyncAPIResource):
                 {
                     "prompt": prompt,
                     "agent": agent,
+                    "detect_elements": detect_elements,
                     "highlight_elements": highlight_elements,
+                    "human_intervention": human_intervention,
+                    "max_steps": max_steps,
                     "model": model,
                     "output_schema": output_schema,
                     "provider": provider,
+                    "secret_values": secret_values,
                     "url": url,
                 },
                 tool_perform_web_task_params.ToolPerformWebTaskParams,

{anchorbrowser-0.3.2 → anchorbrowser-0.3.4}/src/anchorbrowser/types/tool_perform_web_task_params.py

@@ -2,6 +2,7 @@
 
 from __future__ import annotations
 
+from typing import Dict
 from typing_extensions import Literal, Required, Annotated, TypedDict
 
 from .._utils import PropertyInfo
@@ -23,9 +24,27 @@ class ToolPerformWebTaskParams(TypedDict, total=False):
     agent: Literal["browser-use", "openai-cua"]
     """The AI agent to use for task completion. Defaults to browser-use."""
 
+    detect_elements: bool
+    """Enable element detection for better interaction accuracy.
+
+    Improves the agent's ability to identify and interact with UI elements.
+    """
+
     highlight_elements: bool
     """Whether to highlight elements during task execution for better visibility."""
 
+    human_intervention: bool
+    """Allow human intervention during task execution.
+
+    When enabled, the agent can request human input for ambiguous situations.
+    """
+
+    max_steps: int
+    """Maximum number of steps the agent can take to complete the task.
+
+    Defaults to 25.
+    """
+
     model: str
     """The specific model to use for task completion.
 
@@ -39,6 +58,12 @@ class ToolPerformWebTaskParams(TypedDict, total=False):
     provider: Literal["openai", "gemini", "groq", "azure", "xai"]
     """The AI provider to use for task completion."""
 
+    secret_values: Dict[str, str]
+    """Secret values to pass to the agent for secure credential handling.
+
+    Keys and values are passed as environment variables to the agent.
+    """
+
     url: str
     """The URL of the webpage.
 

{anchorbrowser-0.3.2 → anchorbrowser-0.3.4}/tests/api_resources/test_tools.py

@@ -84,10 +84,14 @@ class TestTools:
             prompt="prompt",
             session_id="sessionId",
             agent="browser-use",
+            detect_elements=True,
             highlight_elements=True,
+            human_intervention=True,
+            max_steps=0,
             model="model",
             output_schema={},
             provider="openai",
+            secret_values={"foo": "string"},
             url="url",
         )
         assert_matches_type(ToolPerformWebTaskResponse, tool, path=["response"])
@@ -237,10 +241,14 @@ class TestAsyncTools:
             prompt="prompt",
             session_id="sessionId",
             agent="browser-use",
+            detect_elements=True,
             highlight_elements=True,
+            human_intervention=True,
+            max_steps=0,
             model="model",
             output_schema={},
             provider="openai",
+            secret_values={"foo": "string"},
             url="url",
         )
         assert_matches_type(ToolPerformWebTaskResponse, tool, path=["response"])