Merge origin/main into feature/mongodb-atlas-memory-tool

- Resolve conflicts in README.md by merging MongoDB Atlas Memory Tool and Retrieve Tool sections
- Resolve conflicts in pyproject.toml by including both elasticsearch_memory and mongodb_memory dependencies

Author: Ayan Ray

README.md

@@ -114,7 +114,7 @@ Below is a comprehensive table of all available tools, how to use them with an a
 | calculator | `agent.tool.calculator(expression="2 * sin(pi/4) + log(e**2)")` | Performing mathematical operations, symbolic math, equation solving |
 | code_interpreter | `code_interpreter = AgentCoreCodeInterpreter(region="us-west-2"); agent = Agent(tools=[code_interpreter.code_interpreter])` | Execute code in isolated sandbox environments with multi-language support (Python, JavaScript, TypeScript), persistent sessions, and file operations |
 | use_aws | `agent.tool.use_aws(service_name="s3", operation_name="list_buckets", parameters={}, region="us-west-2")` | Interacting with AWS services, cloud resource management |
-| retrieve | `agent.tool.retrieve(text="What is STRANDS?")` | Retrieving information from Amazon Bedrock Knowledge Bases |
+| retrieve | `agent.tool.retrieve(text="What is STRANDS?")` | Retrieving information from Amazon Bedrock Knowledge Bases with optional metadata |
 | nova_reels | `agent.tool.nova_reels(action="create", text="A cinematic shot of mountains", s3_bucket="my-bucket")` | Create high-quality videos using Amazon Bedrock Nova Reel with configurable parameters via environment variables |
 | agent_core_memory | `agent.tool.agent_core_memory(action="record", content="Hello, I like vegetarian food")` | Store and retrieve memories with Amazon Bedrock Agent Core Memory service |
 | mem0_memory | `agent.tool.mem0_memory(action="store", content="Remember I like to play tennis", user_id="alex")` | Store user and agent memories across agent runs to provide personalized experience |
@@ -504,6 +504,33 @@ result = agent.tool.use_aws(
 )
 ```
 
+### Retrieve Tool
+
+```python
+from strands import Agent
+from strands_tools import retrieve
+
+agent = Agent(tools=[retrieve])
+
+# Basic retrieval without metadata
+result = agent.tool.retrieve(
+    text="What is artificial intelligence?"
+)
+
+# Retrieval with metadata enabled
+result = agent.tool.retrieve(
+    text="What are the latest developments in machine learning?",
+    enableMetadata=True
+)
+
+# Using environment variable to set default metadata behavior
+# Set RETRIEVE_ENABLE_METADATA_DEFAULT=true in your environment
+result = agent.tool.retrieve(
+    text="Tell me about cloud computing"
+    # enableMetadata will default to the environment variable value
+)
+```
+
 ### Batch Tool
 
 ```python
@@ -1099,6 +1126,7 @@ The Mem0 Memory Tool supports three different backend configurations:
 | EDITOR_DIR_TREE_MAX_DEPTH | Maximum depth for directory tree visualization | 2 |
 | EDITOR_DEFAULT_STYLE | Default style for output panels | default |
 | EDITOR_DEFAULT_LANGUAGE | Default language for syntax highlighting | python |
+| EDITOR_DISABLE_BACKUP | Skip creating .bak backup files during edit operations | false |
 
 #### Environment Tool
 
@@ -1145,6 +1173,12 @@ The Mem0 Memory Tool supports three different backend configurations:
 | STRANDS_RSS_UPDATE_INTERVAL | Default amount of time between updating rss feeds in minutes | 60 |
 | STRANDS_RSS_STORAGE_PATH | Default storage path where rss feeds are stored locally | strands_rss_feeds (this may vary based on your system) |
 
+#### Retrieve Tool
+
+| Environment Variable | Description | Default |
+|----------------------|-------------|---------|
+| RETRIEVE_ENABLE_METADATA_DEFAULT | Default setting for enabling metadata in retrieve tool responses | false |
+>>>>>>> origin/main
 #### MongoDB Atlas Memory Tool
 
 | Environment Variable | Description | Default |
@@ -1155,6 +1189,19 @@ The Mem0 Memory Tool supports three different backend configurations:
 | MONGODB_NAMESPACE | Default namespace for memory isolation | default |
 | MONGODB_EMBEDDING_MODEL | Bedrock model for generating embeddings | amazon.titan-embed-text-v2:0 |
 
+#### Retrieve Tool
+
+| Environment Variable | Description | Default |
+|----------------------|-------------|---------|
+| RETRIEVE_ENABLE_METADATA_DEFAULT | Default setting for enabling metadata in retrieve tool responses | false |
+=======
+#### Retrieve Tool
+
+| Environment Variable | Description | Default |
+|----------------------|-------------|---------|
+| RETRIEVE_ENABLE_METADATA_DEFAULT | Default setting for enabling metadata in retrieve tool responses | false |
+>>>>>>> origin/main
+
 #### Video Tools
 
 | Environment Variable | Description | Default | 

pyproject.toml

@@ -38,6 +38,7 @@ dependencies = [
     "markdownify>=1.0.0,<2.0.0",
     "requests>=2.28.0,<3.0.0",
     "aiohttp>=3.8.0,<4.0.0",
+    "typing_extensions>=4.0.0,<5.0.0",
     # Note: Always want the latest tzdata
     "tzdata ; platform_system == 'Windows'",
     "botocore>=1.39.7,<2.0.0",

src/strands_tools/browser/agent_core_browser.py

@@ -10,6 +10,7 @@
 
 from bedrock_agentcore.tools.browser_client import BrowserClient as AgentCoreBrowserClient
 from playwright.async_api import Browser as PlaywrightBrowser
+from typing_extensions import override
 
 from ..utils.aws_util import resolve_region
 from .browser import Browser
@@ -59,6 +60,34 @@ async def create_browser_session(self) -> PlaywrightBrowser:
 
         return browser
 
+    @override
+    async def _setup_session_from_browser(self, browser_or_context):
+        """Setup session for AgentCoreBrowser using existing CDP context.
+
+        AgentCoreBrowser connects via CDP and uses the default context
+        that comes with the connection, avoiding the need to create a new one.
+
+        Args:
+            browser_or_context: PlaywrightBrowser from CDP connection
+
+        Returns:
+            Tuple of (browser, context, page)
+        """
+        session_browser = browser_or_context
+
+        # CDP connections should have a default context
+        if not session_browser.contexts:
+            raise RuntimeError(
+                "AgentCoreBrowser CDP connection has no contexts. "
+                "This may indicate a connection issue with the remote browser."
+            )
+
+        # Use the existing default context from CDP connection
+        session_context = session_browser.contexts[0]
+        session_page = await session_context.new_page()
+
+        return session_browser, session_context, session_page
+
     def close_platform(self) -> None:
         for client in self._client_dict.values():
             try:

src/strands_tools/browser/browser.py

@@ -221,6 +221,31 @@ async def create_browser_session(self) -> PlaywrightBrowser:
         """
         ...
 
+    async def _setup_session_from_browser(self, browser_or_context):
+        """Setup session components from browser or context.
+
+        This method can be overridden by subclasses to customize how
+        browser, context, and page are extracted from the session object.
+
+        Args:
+            browser_or_context: The object returned by create_browser_session()
+
+        Returns:
+            Tuple of (browser, context, page)
+        """
+        if isinstance(browser_or_context, PlaywrightBrowser):
+            # Normal non-persistent case
+            session_browser = browser_or_context
+            session_context = await session_browser.new_context()
+            session_page = await session_context.new_page()
+        else:
+            # Persistent context case
+            session_context = browser_or_context
+            session_browser = session_context.browser
+            session_page = await session_context.new_page()
+
+        return session_browser, session_context, session_page
+
     # Session Management Methods
     def init_session(self, action: InitSessionAction) -> Dict[str, Any]:
         """Initialize a new browser session."""
@@ -238,19 +263,10 @@ async def _async_init_session(self, action: InitSessionAction) -> Dict[str, Any]
 
         try:
             # Create new browser instance for this session
-            session = await self.create_browser_session()
+            browser_or_context = await self.create_browser_session()
 
-            if isinstance(session, PlaywrightBrowser):
-                # Normal non-persistent case
-                session_browser = session
-                session_context = await session_browser.new_context()
-                session_page = await session_context.new_page()
-
-            else:
-                # Persistent context case
-                session_context = session
-                session_browser = session_context.browser
-                session_page = await session_context.new_page()
+            # Let subclasses customize how to extract browser, context, and page
+            session_browser, session_context, session_page = await self._setup_session_from_browser(browser_or_context)
 
             # Create and store session object
             session = BrowserSession(

src/strands_tools/editor.py

@@ -536,8 +536,10 @@ def editor(
 
             # Make replacements and backup
             new_content = content.replace(old_str, new_str)
-            backup_path = f"{path}.bak"
-            shutil.copy2(path, backup_path)
+            disable_backup = os.environ.get("EDITOR_DISABLE_BACKUP", "").lower() == "true"
+            if not disable_backup:
+                backup_path = f"{path}.bak"
+                shutil.copy2(path, backup_path)
 
             # Write new content and update cache
             with open(path, "w") as f:
@@ -606,8 +608,12 @@ def editor(
 
             # Make replacements and backup
             new_content = regex.sub(new_str, content)
-            backup_path = f"{path}.bak"
-            shutil.copy2(path, backup_path)
+            disable_backup = os.environ.get("EDITOR_DISABLE_BACKUP", "").lower() == "true"
+            if not disable_backup:
+                backup_path = f"{path}.bak"
+                shutil.copy2(path, backup_path)
+            else:
+                backup_path = "Disabled"
 
             # Write new content and update cache
             with open(path, "w") as f:
@@ -678,8 +684,10 @@ def editor(
                 raise ValueError(f"insert_line {insert_line} is out of range")
 
             # Make backup
-            backup_path = f"{path}.bak"
-            shutil.copy2(path, backup_path)
+            disable_backup = os.environ.get("EDITOR_DISABLE_BACKUP", "").lower() == "true"
+            if not disable_backup:
+                backup_path = f"{path}.bak"
+                shutil.copy2(path, backup_path)
 
             # Insert and write
             lines.insert(insert_line, new_str)