mirror of
https://github.com/microsoft/agent-framework.git
synced 2026-06-16 21:04:09 +08:00
Giles Odigwe
5affc9c333
* Reorganize A2A samples: client demos in 02-agents, use package A2AExecutor - Move client samples (agent_with_a2a, a2a_agent_as_function_tools) to samples/02-agents/a2a/ - Add new concept samples: polling, stream reconnection, protocol selection - Replace sample agent_executor.py with package-level A2AExecutor (stream=True) - Update 04-hosting/a2a to focus on server-side, point to 02-agents for clients - Add README.md for the new 02-agents/a2a/ sample collection Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Fix streaming artifact coalescing and address PR review feedback A2AExecutor fix: - Generate a stable artifact_id per stream in _run_stream so all streaming chunks share the same ID, enabling proper append=True coalescing per the A2A spec (TaskArtifactUpdateEvent with same artifactId). - Previously, item.message_id was None for OpenAI/Foundry streaming updates, causing the SDK to generate a new random UUID per token (100+ separate artifacts instead of 1 appended artifact). Sample improvements: - Replace join workaround with response.text now that coalescing works - Add background=True to stream reconnection resume call (required for continuation token emission on in-progress tasks) - Fix type ignore specificity in polling sample Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> --------- Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
55 lines
1.9 KiB
Markdown
55 lines
1.9 KiB
Markdown
# A2A Client Samples
|
|
|
|
These samples demonstrate how to **consume** remote A2A-compliant agents using the Agent Framework's `A2AAgent` class.
|
|
|
|
For hosting your own agents as A2A servers, see [`samples/04-hosting/a2a/`](../../04-hosting/a2a/).
|
|
|
|
## Samples
|
|
|
|
| Sample | Concept |
|
|
|--------|---------|
|
|
| [`agent_with_a2a.py`](agent_with_a2a.py) | Basic consumption — non-streaming and streaming |
|
|
| [`a2a_agent_as_function_tools.py`](a2a_agent_as_function_tools.py) | Expose A2A skills as function tools for a host agent |
|
|
| [`a2a_polling.py`](a2a_polling.py) | Poll a long-running task with continuation tokens |
|
|
| [`a2a_stream_reconnection.py`](a2a_stream_reconnection.py) | Resume an interrupted stream via continuation token |
|
|
| [`a2a_protocol_selection.py`](a2a_protocol_selection.py) | Configure preferred protocol bindings (JSONRPC, GRPC, HTTP+JSON) |
|
|
|
|
## Prerequisites
|
|
|
|
- A running A2A-compliant agent server (see `samples/04-hosting/a2a/` to start one)
|
|
- Set `A2A_AGENT_HOST` environment variable to the server URL
|
|
- For `a2a_agent_as_function_tools.py`: also set `FOUNDRY_PROJECT_ENDPOINT` and `FOUNDRY_MODEL`
|
|
|
|
## Running
|
|
|
|
```bash
|
|
cd python/samples/02-agents/a2a
|
|
|
|
# Start an A2A server in another terminal first:
|
|
# cd python/samples/04-hosting/a2a && uv run python a2a_server.py
|
|
|
|
export A2A_AGENT_HOST="http://localhost:5001/"
|
|
uv run python agent_with_a2a.py
|
|
```
|
|
|
|
## Key APIs
|
|
|
|
```python
|
|
from agent_framework.a2a import A2AAgent
|
|
|
|
# Connect to a remote agent
|
|
async with A2AAgent(url="http://localhost:5001/", agent_card=card) as agent:
|
|
# Non-streaming
|
|
response = await agent.run("Hello")
|
|
|
|
# Streaming
|
|
stream = agent.run("Hello", stream=True)
|
|
async for update in stream:
|
|
print(update.text)
|
|
|
|
# Background + polling
|
|
response = await agent.run("Long task", background=True)
|
|
while response.continuation_token:
|
|
response = await agent.poll_task(response.continuation_token)
|
|
```
|