Add Google GenAI plugin samples by brianstrauch · Pull Request #319 · temporalio/samples-python

brianstrauch · 2026-06-18T18:44:59Z

Adds a google_genai_plugin/ sample suite for temporalio.contrib.google_genai, mirroring the strands_plugin/ layout (one feature per sub-directory, each with workflow.py / run_worker.py / run_workflow.py / README.md).

Samples

Sample	Feature	Automated test
`hello_world`	`generate_content`	✅
`tools`	automatic function calling (`activity_as_tool` + plain workflow-method tool)	✅
`streaming`	`generate_content_stream` + `streaming_topic`/`WorkflowStream`	✅
`chat`	multi-turn `client.chats`	✅
`structured_output`	`response_schema` + Pydantic	✅
`mcp`	`TemporalMcpClientSession` with a local echo MCP server	✅
`files`	`client.files.upload`	runnable-only (live API)
`interactions`	`client.interactions` stateful API	runnable-only (live API)
`agents`	`client.agents` CRUD	runnable-only (live API)
`vertex_ai`	`vertexai=True` configuration	runnable-only (GCP creds)

Tests

tests/google_genai_plugin/ uses the plugin's GeminiTestServer to run the model-layer samples offline (no API key); the mcp test additionally registers a real echo MCP server. The files/interactions/agents/vertex_ai samples talk to backends GeminiTestServer does not mock, so they are runnable-only and documented as requiring live credentials.

Registration

pyproject.toml: google-genai dependency group (temporalio[google-genai,pydantic] + mcp) and wheel package
root README.md sample listing
.github/CODEOWNERS

Verification

uv run pytest tests/google_genai_plugin/ → 6 passed
mypy --check-untyped-defs --namespace-packages → no issues
ruff format --check + ruff check --select I → clean
all 10 sample modules import cleanly

Note: the google-genai extra is not yet in a released temporalio; until then install the SDK from source: uv pip install -e "../sdk-python[google-genai,pydantic]".

🤖 Generated with Claude Code

Add a google_genai_plugin/ sample suite for temporalio.contrib.google_genai, mirroring the strands_plugin/ layout (one feature per sub-directory, each with workflow.py / run_worker.py / run_workflow.py / README.md). Samples cover every major plugin feature: - hello_world: generate_content - tools: automatic function calling (activity_as_tool + plain workflow-method tool) - streaming: generate_content_stream + streaming_topic/WorkflowStream - chat: multi-turn client.chats - structured_output: response_schema + Pydantic - mcp: TemporalMcpClientSession with a local echo MCP server - files: client.files.upload (live API) - interactions: client.interactions stateful API (live API) - agents: client.agents CRUD (live API) - vertex_ai: vertexai=True configuration (GCP credentials) Tests under tests/google_genai_plugin/ use the plugin's GeminiTestServer to run the model-layer samples offline; the mcp test additionally registers a real echo MCP server. files/interactions/agents/vertex_ai are runnable-only (require live credentials) and documented as such. Register the suite in pyproject.toml (google-genai dependency group + wheel package), the root README, and CODEOWNERS. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Wrap the workflow.py, run_worker.py, and run_workflow.py bodies of each sample in @@@SNIPSTART/@@@SNIPEND markers (python-google-genai-<sample>-<part>) so the code can be embedded in docs, matching the strands_plugin convention. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Resolve conflicts in .github/CODEOWNERS, README.md, and pyproject.toml by keeping both the google-genai-plugin and incoming AI SDK sample entries in alphabetical order, and regenerate uv.lock. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

The stream publishes Pydantic GenerateContentResponse chunks, so the consumer needs the Pydantic data converter to decode them. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Bump every temporalio requirement to >=1.31.0. The 1.31 google-adk extra requires google-adk 2.x, so widen that pin too, and relax the interactions sample's typing since create/get now return a union with the streaming response type. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

…o google-genai-plugin-samples # Conflicts: # .github/CODEOWNERS # README.md # pyproject.toml # uv.lock

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

The google-genai and strands-agents extras of temporalio shipped in 1.31, which pyproject.toml already requires, so `uv sync --group ...` is enough. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

DABH · 2026-08-04T05:22:04Z

PR description is stale now -

It says the google-genai extra "is not yet in a released temporalio; until then install the SDK from source." That's no longer true: the PR itself pins temporalio>=1.31.0 (released 2026-07-29), the lock resolves temporalio[google-genai,pydantic] from PyPI, and CI passes without any source install. Delete the note.
It says the suite is google_genai_plugin/, but the directory is google_genai/. The name itself is fine (matches the contrib module name, and google_adk_agents already breaks the *_plugin pattern), but the body should match reality.

DABH · 2026-08-04T05:23:21Z

+            ),
+        )
+        recipe = response.parsed
+        assert isinstance(recipe, Recipe)


If Gemini returns something response.parsed can't produce (None on malformed JSON), the AssertionError is a workflow task failure, so the workflow retries the task indefinitely instead of failing visibly. Raise temporalio.exceptions.ApplicationError (arguably non-retryable) with a clear message instead

DABH · 2026-08-04T05:24:02Z

+        )
+        return response.text or ""
+
+    async def recommend_activity(self, weather: str) -> str:


Activity is overloaded term :) Say "pastime" or something instead of "activity" for function name? :)

DABH · 2026-08-04T05:25:00Z

+
+    # Subscribe to the "gemini" topic and print chunks as the model produces them.
+    stream = WorkflowStreamClient.create(client, workflow_id)
+    async for item in stream.subscribe(


We only exit the subscribe loop when a chunk carries finish_reason - if generation errors mid-stream the consumer hangs; and the workflow itself waits forever on the finish signal if the subscriber dies before signaling. For a sample this is tolerable, but a workflow.wait_condition(..., timeout=...) (or at least a README sentence or comment about it) would model the production-grade pattern

DABH · 2026-08-04T05:26:22Z

+
+## What This Sample Demonstrates
+
+- `client.agents.create(id=..., system_instruction=...)`


client.agents.create(id=fixed_id) is non-idempotent: if the create activity succeeds but the completion is lost, the retry will hit "already exists"; and if get/list raises, the agent leaks because delete never runs. A try/finally around the delete - or a short README note on idempotent creates under retries - would teach the right habit in a suite whose whole pitch is durable execution.

DABH · 2026-08-04T05:28:01Z

+        index["i"] += 1
+        return SdkHttpResponse(headers={"content-type": "application/json"}, body=body)
+
+    genai_client._api_client.async_request = fake_async_request  # type: ignore[assignment]


Worth filing an issue asking for GeminiTestServer.plugin(mcp_servers=...) so this helper using a private internal API can be deleted

DABH · 2026-08-04T05:29:08Z

@@ -0,0 +1,27 @@
+"""Start the agents workflow."""
+
+# @@@SNIPSTART python-google-genai-agents-run-workflow


The snip start and end markers wrap entire files right now. Our snipsync convention typically prefers the smallest meaningful unit for workflow/activity snippets so docs readers don't wade through imports. Full-file regions are defensible for the worker/starter scripts (they're meant to be runnable), but consider tightening the workflow-class markers? Just a nit.

DABH · 2026-08-04T05:29:40Z

 * [external_storage_redis](external_storage_redis) - Redis driver for external storage
 * [gevent_async](gevent_async) - Combine gevent and Temporal.
 * [google_adk_agents](google_adk_agents) - Run Google ADK agents as durable Temporal workflows (model calls, tools, multi-agent, MCP, streaming).
+* [google_genai](google_genai) - Run the Google Gemini SDK inside durable Temporal workflows (generate_content, tools/AFC, streaming, chat, structured output, MCP, files, interactions, agents, Vertex AI).


Trim this down. They can click to see the genai samples readme to see the full list of 10 things enumerated.

brianstrauch requested review from a team as code owners June 18, 2026 18:45

brianstrauch and others added 8 commits June 18, 2026 12:07

Use Pydantic data converter in streaming run_workflow

61ae58e

The stream publishes Pydantic GenerateContentResponse chunks, so the consumer needs the Pydantic data converter to decode them. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Merge branch 'main' into google-genai-plugin-samples

8fef9db

Merge remote-tracking branch 'origin/google-genai-plugin-samples' int…

49b6af1

…o google-genai-plugin-samples # Conflicts: # .github/CODEOWNERS # README.md # pyproject.toml # uv.lock

Rename google_genai_plugin to google_genai

771edf2

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

Remove stale unreleased-extra notes from READMEs

d703df4

The google-genai and strands-agents extras of temporalio shipped in 1.31, which pyproject.toml already requires, so `uv sync --group ...` is enough. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

brianstrauch requested a review from a team as a code owner July 29, 2026 19:14

brianstrauch mentioned this pull request Jul 29, 2026

Add Python Google GenAI plugin integration guide temporalio/documentation#4991

Open

brianstrauch enabled auto-merge (squash) July 29, 2026 19:27

DABH reviewed Aug 4, 2026

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Add Google GenAI plugin samples - #319

Add Google GenAI plugin samples#319
brianstrauch wants to merge 9 commits into
mainfrom
google-genai-plugin-samples

brianstrauch commented Jun 18, 2026

Uh oh!

DABH commented Aug 4, 2026

Uh oh!

DABH Aug 4, 2026

Uh oh!

DABH Aug 4, 2026

Uh oh!

DABH Aug 4, 2026

Uh oh!

DABH Aug 4, 2026

Uh oh!

DABH Aug 4, 2026

Uh oh!

DABH Aug 4, 2026

Uh oh!

DABH Aug 4, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants


		## What This Sample Demonstrates

		- `client.agents.create(id=..., system_instruction=...)`

		@@ -0,0 +1,27 @@
		"""Start the agents workflow."""

		# @@@SNIPSTART python-google-genai-agents-run-workflow

Uh oh!

Conversation

brianstrauch commented Jun 18, 2026

Samples

Tests

Registration

Verification

Uh oh!

DABH commented Aug 4, 2026

Uh oh!

DABH Aug 4, 2026

Choose a reason for hiding this comment

Uh oh!

DABH Aug 4, 2026

Choose a reason for hiding this comment

Uh oh!

DABH Aug 4, 2026

Choose a reason for hiding this comment

Uh oh!

DABH Aug 4, 2026

Choose a reason for hiding this comment

Uh oh!

DABH Aug 4, 2026

Choose a reason for hiding this comment

Uh oh!

DABH Aug 4, 2026

Choose a reason for hiding this comment

Uh oh!

DABH Aug 4, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants