a2a-adapter 0.1.3__tar.gz → 0.1.4__tar.gz

{a2a_adapter-0.1.3 → a2a_adapter-0.1.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: a2a-adapter
-Version: 0.1.3
+Version: 0.1.4
 Summary: A2A Protocol Adapter SDK for integrating various agent frameworks
 Author-email: HYBRO AI <info@hybro.ai>
 License: Apache-2.0
@@ -52,20 +52,24 @@ Dynamic: license-file
 
 **🚀 Open Source A2A Protocol Adapter SDK - Make Any Agent Framework A2A-Compatible in 3 Lines**
 
-A Python SDK that enables seamless integration of various agent frameworks (n8n, CrewAI, LangChain, etc.) with the [A2A (Agent-to-Agent) Protocol](https://github.com/a2aproject/A2A). Build interoperable AI agent systems that can communicate across different platforms and frameworks.
+A Python SDK that enables seamless integration of various agent frameworks (n8n, LangGraph, CrewAI, LangChain, etc.) with the [A2A (Agent-to-Agent) Protocol](https://github.com/a2aproject/A2A). Build interoperable AI agent systems that can communicate across different platforms and frameworks.
 
 **✨ Key Benefits:**
 
 - 🔌 **3-line setup** - Expose any agent as A2A-compliant
-- 🌐 **Framework agnostic** - Works with n8n, CrewAI, LangChain, and more
+- 🌐 **Framework agnostic** - Works with n8n, LangGraph, CrewAI, LangChain, and more
 - 🌊 **Streaming support** - Built-in streaming for real-time responses
 - 🎯 **Production ready** - Type-safe, well-tested, and actively maintained
 
+**▶️ Demo: n8n → A2A Agent**
+
+[![A2A Adapter Demo](https://img.youtube.com/vi/rHWi7tLQ444/0.jpg)](https://youtu.be/rHWi7tLQ444)
+
 ## Features
 
-✨ **Framework Agnostic**: Integrate n8n workflows, CrewAI crews, LangChain chains, and more
+✨ **Framework Agnostic**: Integrate n8n workflows, LangGraph workflows, CrewAI crews, LangChain chains, and more
 🔌 **Simple API**: 3-line setup to expose any agent as A2A-compliant
-🌊 **Streaming Support**: Built-in streaming for LangChain and custom adapters
+🌊 **Streaming Support**: Built-in streaming for LangGraph, LangChain, and custom adapters
 🎯 **Type Safe**: Leverages official A2A SDK types
 🔧 **Extensible**: Easy to add custom adapters for new frameworks
 📦 **Minimal Dependencies**: Optional dependencies per framework
@@ -81,9 +85,10 @@ A Python SDK that enables seamless integration of various agent frameworks (n8n,
 ┌─────────────────┐
 │  A2A Adapter    │  (This SDK)
 │   - N8n         │
+│   - LangGraph   │
 │   - CrewAI      │
 │   - LangChain   │
-│   - Custom      │
+│   - Callable    │
 └────────┬────────┘
          │
          ▼
@@ -198,6 +203,17 @@ adapter = await load_a2a_agent({
 })
 ```
 
+### LangGraph Workflow → A2A Agent (with Streaming)
+
+```python
+adapter = await load_a2a_agent({
+    "adapter": "langgraph",
+    "graph": your_compiled_graph,
+    "input_key": "messages",
+    "output_key": "output"
+})
+```
+
 ### Custom Function → A2A Agent
 
 ```python
@@ -271,9 +287,35 @@ class StreamingAdapter(BaseAgentAdapter):
         return True
 ```
 
-### Using with LangGraph
+### LangGraph Workflow as A2A Server
 
-Integrate A2A agents into LangGraph workflows:
+Expose a LangGraph workflow as an A2A server:
+
+```python
+from langgraph.graph import StateGraph, END
+
+# Build your workflow
+builder = StateGraph(YourState)
+builder.add_node("process", process_node)
+builder.set_entry_point("process")
+builder.add_edge("process", END)
+graph = builder.compile()
+
+# Expose as A2A agent
+adapter = await load_a2a_agent({
+    "adapter": "langgraph",
+    "graph": graph,
+    "input_key": "messages",
+    "output_key": "output"
+})
+serve_agent(agent_card=card, adapter=adapter, port=9002)
+```
+
+See [examples/07_langgraph_server.py](https://github.com/hybroai/a2a-adapter/blob/main/examples/07_langgraph_server.py) for complete example.
+
+### Using A2A Agents from LangGraph
+
+Call A2A agents from within a LangGraph workflow:
 
 ```python
 from langgraph.graph import StateGraph
@@ -332,6 +374,19 @@ See [examples/06_langgraph_single_agent.py](https://github.com/hybroai/a2a-adapt
 }
 ```
 
+### LangGraph Adapter
+
+```python
+{
+    "adapter": "langgraph",
+    "graph": compiled_graph,      # Required: CompiledGraph from StateGraph.compile()
+    "input_key": "messages",      # Optional, default: "messages" (for chat) or "input"
+    "output_key": None,           # Optional, extracts specific key from final state
+    "async_mode": False,          # Optional, enables async task execution
+    "async_timeout": 300          # Optional, timeout for async mode (default: 300s)
+}
+```
+
 ### Callable Adapter
 
 ```python
@@ -351,7 +406,8 @@ The `examples/` directory contains complete working examples:
 - **03_single_langchain_agent.py** - LangChain streaming agent
 - **04_single_agent_client.py** - A2A client for testing
 - **05_custom_adapter.py** - Custom adapter implementations
-- **06_langgraph_single_agent.py** - LangGraph + A2A integration
+- **06_langgraph_single_agent.py** - Calling A2A agents from LangGraph
+- **07_langgraph_server.py** - LangGraph workflow as A2A server
 
 Run any example:
 
@@ -457,11 +513,13 @@ Check if this adapter supports streaming responses.
 
 ## Framework Support
 
-| Framework     | Adapter                 | Non-Streaming | Streaming  | Status     |
-| ------------- | ----------------------- | ------------- | ---------- | ---------- |
-| **n8n**       | `N8nAgentAdapter`       | ✅            | 🔜 Planned | ✅ Stable  |
-| **CrewAI**    | `CrewAIAgentAdapter`    | 🔜 Planned    | 🔜 Planned | 🔜 Planned |
-| **LangChain** | `LangChainAgentAdapter` | 🔜 Planned    | 🔜 Planned | 🔜 Planned |
+| Framework     | Adapter                  | Non-Streaming | Streaming | Async Tasks | Status    |
+| ------------- | ------------------------ | ------------- | --------- | ----------- | --------- |
+| **n8n**       | `N8nAgentAdapter`        | ✅            | ❌        | ✅          | ✅ Stable |
+| **LangGraph** | `LangGraphAgentAdapter`  | ✅            | ✅        | ✅          | ✅ Stable |
+| **CrewAI**    | `CrewAIAgentAdapter`     | ✅            | ❌        | ✅          | ✅ Stable |
+| **LangChain** | `LangChainAgentAdapter`  | ✅            | ✅        | ❌          | ✅ Stable |
+| **Callable**  | `CallableAgentAdapter`   | ✅            | ✅        | ❌          | ✅ Stable |
 
 ## 🤝 Contributing
 
@@ -488,12 +546,13 @@ We welcome contributions from the community! Whether you're fixing bugs, adding
 ## Roadmap
 
 - [x] Core adapter abstraction
-- [x] N8n adapter
-- [ ] CrewAI adapter
-- [ ] LangChain adapter with streaming
-- [ ] Callable adapter
-- [ ] Comprehensive examples
-- [ ] Task support (async execution pattern)
+- [x] N8n adapter (with async task support)
+- [x] LangGraph adapter (with streaming and async tasks)
+- [x] CrewAI adapter (with async task support)
+- [x] LangChain adapter (with streaming)
+- [x] Callable adapter (with streaming)
+- [x] Comprehensive examples
+- [x] Task support (async execution pattern)
 - [ ] Artifact support (file uploads/downloads)
 - [ ] AutoGen adapter
 - [ ] Semantic Kernel adapter
@@ -528,12 +587,6 @@ We welcome contributions from the community! Whether you're fixing bugs, adding
 
 Apache-2.0 License - see [LICENSE](https://github.com/hybroai/a2a-adapter/blob/main/LICENSE) file for details.
 
-## Credits
-
-Built with ❤️ by [HYBRO AI](https://hybro.ai)
-
-Powered by the [A2A Protocol](https://github.com/a2aproject/A2A)
-
 ## 💬 Community & Support
 
 - 📚 **[Full Documentation](https://github.com/hybroai/a2a-adapter/blob/main/README.md)** - Complete API reference and guides

{a2a_adapter-0.1.3 → a2a_adapter-0.1.4}/README.md

@@ -7,20 +7,24 @@
 
 **🚀 Open Source A2A Protocol Adapter SDK - Make Any Agent Framework A2A-Compatible in 3 Lines**
 
-A Python SDK that enables seamless integration of various agent frameworks (n8n, CrewAI, LangChain, etc.) with the [A2A (Agent-to-Agent) Protocol](https://github.com/a2aproject/A2A). Build interoperable AI agent systems that can communicate across different platforms and frameworks.
+A Python SDK that enables seamless integration of various agent frameworks (n8n, LangGraph, CrewAI, LangChain, etc.) with the [A2A (Agent-to-Agent) Protocol](https://github.com/a2aproject/A2A). Build interoperable AI agent systems that can communicate across different platforms and frameworks.
 
 **✨ Key Benefits:**
 
 - 🔌 **3-line setup** - Expose any agent as A2A-compliant
-- 🌐 **Framework agnostic** - Works with n8n, CrewAI, LangChain, and more
+- 🌐 **Framework agnostic** - Works with n8n, LangGraph, CrewAI, LangChain, and more
 - 🌊 **Streaming support** - Built-in streaming for real-time responses
 - 🎯 **Production ready** - Type-safe, well-tested, and actively maintained
 
+**▶️ Demo: n8n → A2A Agent**
+
+[![A2A Adapter Demo](https://img.youtube.com/vi/rHWi7tLQ444/0.jpg)](https://youtu.be/rHWi7tLQ444)
+
 ## Features
 
-✨ **Framework Agnostic**: Integrate n8n workflows, CrewAI crews, LangChain chains, and more
+✨ **Framework Agnostic**: Integrate n8n workflows, LangGraph workflows, CrewAI crews, LangChain chains, and more
 🔌 **Simple API**: 3-line setup to expose any agent as A2A-compliant
-🌊 **Streaming Support**: Built-in streaming for LangChain and custom adapters
+🌊 **Streaming Support**: Built-in streaming for LangGraph, LangChain, and custom adapters
 🎯 **Type Safe**: Leverages official A2A SDK types
 🔧 **Extensible**: Easy to add custom adapters for new frameworks
 📦 **Minimal Dependencies**: Optional dependencies per framework
@@ -36,9 +40,10 @@ A Python SDK that enables seamless integration of various agent frameworks (n8n,
 ┌─────────────────┐
 │  A2A Adapter    │  (This SDK)
 │   - N8n         │
+│   - LangGraph   │
 │   - CrewAI      │
 │   - LangChain   │
-│   - Custom      │
+│   - Callable    │
 └────────┬────────┘
          │
          ▼
@@ -153,6 +158,17 @@ adapter = await load_a2a_agent({
 })
 ```
 
+### LangGraph Workflow → A2A Agent (with Streaming)
+
+```python
+adapter = await load_a2a_agent({
+    "adapter": "langgraph",
+    "graph": your_compiled_graph,
+    "input_key": "messages",
+    "output_key": "output"
+})
+```
+
 ### Custom Function → A2A Agent
 
 ```python
@@ -226,9 +242,35 @@ class StreamingAdapter(BaseAgentAdapter):
         return True
 ```
 
-### Using with LangGraph
+### LangGraph Workflow as A2A Server
 
-Integrate A2A agents into LangGraph workflows:
+Expose a LangGraph workflow as an A2A server:
+
+```python
+from langgraph.graph import StateGraph, END
+
+# Build your workflow
+builder = StateGraph(YourState)
+builder.add_node("process", process_node)
+builder.set_entry_point("process")
+builder.add_edge("process", END)
+graph = builder.compile()
+
+# Expose as A2A agent
+adapter = await load_a2a_agent({
+    "adapter": "langgraph",
+    "graph": graph,
+    "input_key": "messages",
+    "output_key": "output"
+})
+serve_agent(agent_card=card, adapter=adapter, port=9002)
+```
+
+See [examples/07_langgraph_server.py](https://github.com/hybroai/a2a-adapter/blob/main/examples/07_langgraph_server.py) for complete example.
+
+### Using A2A Agents from LangGraph
+
+Call A2A agents from within a LangGraph workflow:
 
 ```python
 from langgraph.graph import StateGraph
@@ -287,6 +329,19 @@ See [examples/06_langgraph_single_agent.py](https://github.com/hybroai/a2a-adapt
 }
 ```
 
+### LangGraph Adapter
+
+```python
+{
+    "adapter": "langgraph",
+    "graph": compiled_graph,      # Required: CompiledGraph from StateGraph.compile()
+    "input_key": "messages",      # Optional, default: "messages" (for chat) or "input"
+    "output_key": None,           # Optional, extracts specific key from final state
+    "async_mode": False,          # Optional, enables async task execution
+    "async_timeout": 300          # Optional, timeout for async mode (default: 300s)
+}
+```
+
 ### Callable Adapter
 
 ```python
@@ -306,7 +361,8 @@ The `examples/` directory contains complete working examples:
 - **03_single_langchain_agent.py** - LangChain streaming agent
 - **04_single_agent_client.py** - A2A client for testing
 - **05_custom_adapter.py** - Custom adapter implementations
-- **06_langgraph_single_agent.py** - LangGraph + A2A integration
+- **06_langgraph_single_agent.py** - Calling A2A agents from LangGraph
+- **07_langgraph_server.py** - LangGraph workflow as A2A server
 
 Run any example:
 
@@ -412,11 +468,13 @@ Check if this adapter supports streaming responses.
 
 ## Framework Support
 
-| Framework     | Adapter                 | Non-Streaming | Streaming  | Status     |
-| ------------- | ----------------------- | ------------- | ---------- | ---------- |
-| **n8n**       | `N8nAgentAdapter`       | ✅            | 🔜 Planned | ✅ Stable  |
-| **CrewAI**    | `CrewAIAgentAdapter`    | 🔜 Planned    | 🔜 Planned | 🔜 Planned |
-| **LangChain** | `LangChainAgentAdapter` | 🔜 Planned    | 🔜 Planned | 🔜 Planned |
+| Framework     | Adapter                  | Non-Streaming | Streaming | Async Tasks | Status    |
+| ------------- | ------------------------ | ------------- | --------- | ----------- | --------- |
+| **n8n**       | `N8nAgentAdapter`        | ✅            | ❌        | ✅          | ✅ Stable |
+| **LangGraph** | `LangGraphAgentAdapter`  | ✅            | ✅        | ✅          | ✅ Stable |
+| **CrewAI**    | `CrewAIAgentAdapter`     | ✅            | ❌        | ✅          | ✅ Stable |
+| **LangChain** | `LangChainAgentAdapter`  | ✅            | ✅        | ❌          | ✅ Stable |
+| **Callable**  | `CallableAgentAdapter`   | ✅            | ✅        | ❌          | ✅ Stable |
 
 ## 🤝 Contributing
 
@@ -443,12 +501,13 @@ We welcome contributions from the community! Whether you're fixing bugs, adding
 ## Roadmap
 
 - [x] Core adapter abstraction
-- [x] N8n adapter
-- [ ] CrewAI adapter
-- [ ] LangChain adapter with streaming
-- [ ] Callable adapter
-- [ ] Comprehensive examples
-- [ ] Task support (async execution pattern)
+- [x] N8n adapter (with async task support)
+- [x] LangGraph adapter (with streaming and async tasks)
+- [x] CrewAI adapter (with async task support)
+- [x] LangChain adapter (with streaming)
+- [x] Callable adapter (with streaming)
+- [x] Comprehensive examples
+- [x] Task support (async execution pattern)
 - [ ] Artifact support (file uploads/downloads)
 - [ ] AutoGen adapter
 - [ ] Semantic Kernel adapter
@@ -483,12 +542,6 @@ We welcome contributions from the community! Whether you're fixing bugs, adding
 
 Apache-2.0 License - see [LICENSE](https://github.com/hybroai/a2a-adapter/blob/main/LICENSE) file for details.
 
-## Credits
-
-Built with ❤️ by [HYBRO AI](https://hybro.ai)
-
-Powered by the [A2A Protocol](https://github.com/a2aproject/A2A)
-
 ## 💬 Community & Support
 
 - 📚 **[Full Documentation](https://github.com/hybroai/a2a-adapter/blob/main/README.md)** - Complete API reference and guides

{a2a_adapter-0.1.3 → a2a_adapter-0.1.4}/a2a_adapter/__init__.py

@@ -26,7 +26,7 @@ Example:
     >>> asyncio.run(main())
 """
 
-__version__ = "0.1.3"
+__version__ = "0.1.4"
 
 from .adapter import BaseAgentAdapter
 from .client import build_agent_app, serve_agent

{a2a_adapter-0.1.3 → a2a_adapter-0.1.4}/a2a_adapter/integrations/__init__.py

@@ -5,6 +5,7 @@ This package contains concrete adapter implementations for various agent framewo
 - n8n: HTTP webhook-based workflows
 - CrewAI: Multi-agent collaboration framework
 - LangChain: LLM application framework with LCEL support
+- LangGraph: Stateful workflow orchestration framework
 - Callable: Generic Python async function adapter
 """
 
@@ -12,6 +13,7 @@ __all__ = [
     "N8nAgentAdapter",
     "CrewAIAgentAdapter",
     "LangChainAgentAdapter",
+    "LangGraphAgentAdapter",
     "CallableAgentAdapter",
 ]
 
@@ -26,8 +28,10 @@ def __getattr__(name: str):
     elif name == "LangChainAgentAdapter":
         from .langchain import LangChainAgentAdapter
         return LangChainAgentAdapter
+    elif name == "LangGraphAgentAdapter":
+        from .langgraph import LangGraphAgentAdapter
+        return LangGraphAgentAdapter
     elif name == "CallableAgentAdapter":
         from .callable import CallableAgentAdapter
         return CallableAgentAdapter
     raise AttributeError(f"module {__name__!r} has no attribute {name!r}")
-

a2a_adapter-0.1.4/a2a_adapter/integrations/callable.py

@@ -0,0 +1,286 @@
+"""
+Generic callable adapter for A2A Protocol.
+
+This adapter allows any async Python function to be exposed as an A2A-compliant
+agent, providing maximum flexibility for custom implementations.
+"""
+
+import json
+import logging
+import uuid
+from typing import Any, AsyncIterator, Callable, Dict
+
+from a2a.types import (
+    Message,
+    MessageSendParams,
+    Task,
+    TextPart,
+    Role,
+    Part,
+)
+from ..adapter import BaseAgentAdapter
+
+logger = logging.getLogger(__name__)
+
+
+class CallableAgentAdapter(BaseAgentAdapter):
+    """
+    Adapter for integrating custom async functions as A2A agents.
+
+    This adapter wraps any async callable (function, coroutine) and handles
+    the A2A protocol translation. The callable should accept a dictionary
+    input and return either a string or dictionary output.
+
+    For streaming support, the callable should be an async generator that
+    yields string chunks.
+
+    Example (non-streaming):
+        >>> async def my_agent(inputs: dict) -> str:
+        ...     message = inputs["message"]
+        ...     return f"Processed: {message}"
+        >>>
+        >>> adapter = CallableAgentAdapter(func=my_agent)
+
+    Example (streaming):
+        >>> async def my_streaming_agent(inputs: dict):
+        ...     message = inputs["message"]
+        ...     for word in message.split():
+        ...         yield word + " "
+        >>>
+        >>> adapter = CallableAgentAdapter(
+        ...     func=my_streaming_agent,
+        ...     supports_streaming=True
+        ... )
+    """
+
+    def __init__(
+        self,
+        func: Callable,
+        supports_streaming: bool = False,
+    ):
+        """
+        Initialize the callable adapter.
+
+        Args:
+            func: An async callable that processes the agent logic.
+                  For non-streaming: Should accept Dict[str, Any] and return str or Dict.
+                  For streaming: Should be an async generator yielding str chunks.
+            supports_streaming: Whether the function supports streaming (default: False)
+        """
+        self.func = func
+        self._supports_streaming = supports_streaming
+
+    # ---------- Input mapping ----------
+
+    async def to_framework(self, params: MessageSendParams) -> Dict[str, Any]:
+        """
+        Convert A2A message parameters to a dictionary for the callable.
+
+        Extracts the user's message and relevant metadata.
+
+        Args:
+            params: A2A message parameters
+
+        Returns:
+            Dictionary with input data for the callable
+        """
+        user_message = ""
+
+        # Extract message from A2A params (new format with message.parts)
+        if hasattr(params, "message") and params.message:
+            msg = params.message
+            if hasattr(msg, "parts") and msg.parts:
+                text_parts = []
+                for part in msg.parts:
+                    # Handle Part(root=TextPart(...)) structure
+                    if hasattr(part, "root") and hasattr(part.root, "text"):
+                        text_parts.append(part.root.text)
+                    # Handle direct TextPart
+                    elif hasattr(part, "text"):
+                        text_parts.append(part.text)
+                user_message = self._join_text_parts(text_parts)
+
+        # Legacy support for messages array (deprecated)
+        elif getattr(params, "messages", None):
+            last = params.messages[-1]
+            content = getattr(last, "content", "")
+            if isinstance(content, str):
+                user_message = content.strip()
+            elif isinstance(content, list):
+                text_parts = []
+                for item in content:
+                    txt = getattr(item, "text", None)
+                    if txt and isinstance(txt, str) and txt.strip():
+                        text_parts.append(txt.strip())
+                user_message = self._join_text_parts(text_parts)
+
+        # Extract metadata
+        context_id = self._extract_context_id(params)
+
+        # Build input dictionary with useful fields
+        return {
+            "message": user_message,
+            "context_id": context_id,
+            "params": params,  # Full params for advanced use cases
+        }
+
+    @staticmethod
+    def _join_text_parts(parts: list[str]) -> str:
+        """Join text parts into a single string."""
+        if not parts:
+            return ""
+        text = " ".join(p.strip() for p in parts if p)
+        return text.strip()
+
+    def _extract_context_id(self, params: MessageSendParams) -> str | None:
+        """Extract context_id from MessageSendParams."""
+        if hasattr(params, "message") and params.message:
+            return getattr(params.message, "context_id", None)
+        return None
+
+    # ---------- Framework call ----------
+
+    async def call_framework(
+        self, framework_input: Dict[str, Any], params: MessageSendParams
+    ) -> Any:
+        """
+        Execute the callable function with the provided input.
+
+        Args:
+            framework_input: Input dictionary for the function
+            params: Original A2A parameters (for context)
+
+        Returns:
+            Function execution output
+
+        Raises:
+            Exception: If function execution fails
+        """
+        logger.debug("Invoking callable with input keys: %s", list(framework_input.keys()))
+        result = await self.func(framework_input)
+        logger.debug("Callable returned: %s", type(result).__name__)
+        return result
+
+    # ---------- Output mapping ----------
+
+    async def from_framework(
+        self, framework_output: Any, params: MessageSendParams
+    ) -> Message | Task:
+        """
+        Convert callable output to A2A Message.
+
+        Args:
+            framework_output: Output from the callable
+            params: Original A2A parameters
+
+        Returns:
+            A2A Message with the function's response
+        """
+        response_text = self._extract_output_text(framework_output)
+        context_id = self._extract_context_id(params)
+
+        return Message(
+            role=Role.agent,
+            message_id=str(uuid.uuid4()),
+            context_id=context_id,
+            parts=[Part(root=TextPart(text=response_text))],
+        )
+
+    def _extract_output_text(self, framework_output: Any) -> str:
+        """
+        Extract text content from callable output.
+
+        Args:
+            framework_output: Output from the callable
+
+        Returns:
+            Extracted text string
+        """
+        # Dictionary output
+        if isinstance(framework_output, dict):
+            # Try common output keys
+            for key in ["response", "output", "result", "answer", "text", "message"]:
+                if key in framework_output:
+                    return str(framework_output[key])
+            # Fallback: serialize as JSON
+            return json.dumps(framework_output, indent=2)
+
+        # String or other type - convert to string
+        return str(framework_output)
+
+    # ---------- Streaming support ----------
+
+    async def handle_stream(
+        self, params: MessageSendParams
+    ) -> AsyncIterator[Dict[str, Any]]:
+        """
+        Handle a streaming A2A message request.
+
+        The wrapped function must be an async generator for streaming to work.
+
+        Args:
+            params: A2A message parameters
+
+        Yields:
+            Server-Sent Events compatible dictionaries with streaming chunks
+
+        Raises:
+            NotImplementedError: If streaming is not enabled for this adapter
+        """
+        if not self._supports_streaming:
+            raise NotImplementedError(
+                "CallableAgentAdapter: streaming not enabled for this function. "
+                "Initialize with supports_streaming=True and provide an async generator."
+            )
+
+        framework_input = await self.to_framework(params)
+        context_id = self._extract_context_id(params)
+        message_id = str(uuid.uuid4())
+
+        logger.debug("Starting streaming call")
+
+        accumulated_text = ""
+
+        # Call the async generator function
+        async for chunk in self.func(framework_input):
+            # Convert chunk to string if needed
+            text = str(chunk) if not isinstance(chunk, str) else chunk
+
+            if text:
+                accumulated_text += text
+                # Yield SSE-compatible event
+                yield {
+                    "event": "message",
+                    "data": json.dumps({
+                        "type": "content",
+                        "content": text,
+                    }),
+                }
+
+        # Send final message with complete response
+        final_message = Message(
+            role=Role.agent,
+            message_id=message_id,
+            context_id=context_id,
+            parts=[Part(root=TextPart(text=accumulated_text))],
+        )
+
+        # Send completion event
+        yield {
+            "event": "done",
+            "data": json.dumps({
+                "status": "completed",
+                "message": final_message.model_dump() if hasattr(final_message, "model_dump") else str(final_message),
+            }),
+        }
+
+        logger.debug("Streaming call completed")
+
+    def supports_streaming(self) -> bool:
+        """
+        Check if this adapter supports streaming.
+
+        Returns:
+            True if streaming is enabled, False otherwise
+        """
+        return self._supports_streaming