hybroai a2a-adapter SDK
Convert any AI agent into an A2A Protocol server. Adapters only answer "given text, return text" -- all protocol handling (JSON-RPC, task management, SSE streaming, push notifications) is delegated to the A2A SDK automatically.
Install
pip install a2a-adapter # Core (n8n, callable, ollama, openclaw)
pip install a2a-adapter[crewai] # + CrewAI
pip install a2a-adapter[langchain] # + LangChain
pip install a2a-adapter[langgraph] # + LangGraph
pip install a2a-adapter[all] # Everything
CLI Pattern
Expose an installed local agent without writing Python:
a2a-adapter pi --port 9012
a2a-adapter codex --port 9011
a2a-adapter claude --port 9010
a2a-adapter openclaw --port 9008
Core Pattern (3 lines)
from a2a_adapter import XxxAdapter, serve_agent
adapter = XxxAdapter(...) # Create adapter
serve_agent(adapter, port=9000) # Start A2A server
serve_agent() starts uvicorn with an auto-generated AgentCard at /.well-known/agent-card.json.
Decision Guide
| Scenario | Use |
|---|---|
| Wrap a function | CallableAdapter(func=fn) |
| n8n workflow | N8nAdapter(webhook_url=...) |
| LangChain chain | LangChainAdapter(runnable=chain) |
| LangGraph workflow | LangGraphAdapter(graph=graph) |
| CrewAI crew | CrewAIAdapter(crew=crew) |
| Claude Code CLI | ClaudeCodeAdapter(working_dir=...) |
| Codex CLI | CodexAdapter(working_dir=...) |
| Pi coding agent | PiAdapter(working_dir=...) |
| OpenClaw agent | OpenClawAdapter(...) |
| Hermes Agent | HermesAdapter(...) |
| Local Ollama model | OllamaAdapter(model="llama3.2") |
| Any other framework | Subclass BaseA2AAdapter, implement invoke() |
| Need streaming | Implement stream() or use LangChain/LangGraph/Ollama (auto) |
| Need multimodal output | Return list[Part] from invoke() |
| Production deploy | to_a2a(adapter) -> ASGI server |
| Config-driven | load_adapter({"adapter": "n8n", ...}) |
Custom Adapter Template
from a2a_adapter import BaseA2AAdapter, AdapterMetadata, serve_agent
class MyAdapter(BaseA2AAdapter):
# REQUIRED: the only method you must implement
async def invoke(self, user_input: str, context_id: str | None = None, **kwargs) -> str:
# kwargs['context'] provides the full A2A RequestContext
return await call_my_framework(user_input)
# OPTIONAL: implement for streaming support (auto-detected)
async def stream(self, user_input: str, context_id: str | None = None, **kwargs):
async for chunk in my_framework_stream(user_input):
yield str(chunk)
# OPTIONAL: cancellation support
async def cancel(self, context_id: str | None = None, **kwargs) -> None:
pass
# OPTIONAL: resource cleanup (called by async with)
async def close(self) -> None:
pass
# OPTIONAL: metadata for auto AgentCard generation
def get_metadata(self) -> AdapterMetadata:
return AdapterMetadata(
name="My Agent", description="Does something useful",
version="1.0.0", streaming=True,
skills=[{"id": "main", "name": "Main Skill", "description": "..."}],
)
serve_agent(MyAdapter(), port=9000)
Reference
See api-reference.md in this directory for full adapter parameters, server function signatures, multimodal response patterns, and architecture details.