temporalio.contrib.google_genai.GoogleGenAIPlugin

class documentation

class GoogleGenAIPlugin(SimplePlugin): (source)

Constructor: GoogleGenAIPlugin(client, extra_credentials, mcp_servers, mcp_connection_idle_timeout)

A Temporal Worker Plugin configured for the Google Gemini SDK.

Warning

This class is experimental and may change in future versions. Use with caution in production environments.

This plugin registers the gemini_api_client_async_request activity using the provided genai.Client with real credentials. Workflows construct a temporalio.contrib.google_genai.TemporalAsyncClient to get an AsyncClient backed by a TemporalApiClient that routes all API calls through this activity.

No credentials are passed to or from the workflow. Auth material never appears in Temporal's event history.

Example (Gemini Developer API):

client = genai.Client(api_key=os.environ["GOOGLE_API_KEY"])
plugin = GoogleGenAIPlugin(client)

Example (Vertex AI):

client = genai.Client(
    vertexai=True, project="my-project", location="us-central1",
)
plugin = GoogleGenAIPlugin(client)

Example (with separate GCS credentials for file registration):

client = genai.Client(api_key=os.environ["GOOGLE_API_KEY"])
gcs_creds, _ = google.auth.default()
plugin = GoogleGenAIPlugin(client, extra_credentials=gcs_creds)

Method	`__init__`	Initialize the Gemini plugin.
Instance Variable	`_api_caller`	Undocumented

Inherited from SimplePlugin:

Method	`configure_client`	See base class.
Method	`configure_replayer`	See base class.
Method	`configure_worker`	See base class.
Async Method	`connect_service_client`	See base class.
Method	`name`	See base class.
Async Method	`run_replayer`	See base class.
Async Method	`run_worker`	See base class.
Instance Variable	`activities`	Undocumented
Instance Variable	`data_converter`	Undocumented
Instance Variable	`interceptors`	Undocumented
Instance Variable	`nexus_service_handlers`	Undocumented
Instance Variable	`run_context`	Undocumented
Instance Variable	`workflow_failure_exception_types`	Undocumented
Instance Variable	`workflow_runner`	Undocumented
Instance Variable	`workflows`	Undocumented
Instance Variable	`_name`	Undocumented

def __init__(self, client: GeminiClient, extra_credentials: google.auth.credentials.Credentials | None = None, mcp_servers: dict[str, McpSessionFactory] | None = None, mcp_connection_idle_timeout: timedelta | None = None): (source) ¶

overrides temporalio.plugin.SimplePlugin.__init__

Initialize the Gemini plugin.

Parameters
client:`GeminiClient`	A fully configured `genai.Client` instance. All credential management, HTTP client configuration, etc. is the responsibility of the caller.
extra_credentials:`google.auth.credentials.Credentials \| None`	Optional Google Cloud credentials used for operations that require explicit auth (e.g. `files.register_files()`). If not provided, the client's own credentials are used.
mcp_servers:`dict[str, McpSessionFactory] \| None`	MCP servers to expose to workflows, keyed by name. Each value is a factory returning an async context manager that yields a connected, initialized `mcp.ClientSession`. A workflow references a server by name with `TemporalMcpClientSession(name)` in a `generate_content` `tools` list; `list_tools` / `call_tool` then run as the `{name}-list-tools` / `{name}-call-tool` activities against a worker-side connection. Requires the `mcp` package.
mcp_connection_idle_timeout:`timedelta \| None`	How long a worker-process MCP connection stays open while idle before being disconnected (the timer resets on each reuse). Defaults to 5 minutes.

_api_caller = (source) ¶

Undocumented