[pull] main from google:main

.github/workflows/update-python-docs.yaml

@@ -61,7 +61,7 @@ jobs:
           GITHUB_TOKEN: ${{ secrets.ADK_BOT_GITHUB_TOKEN }}
           GOOGLE_CLOUD_PROJECT: ${{ secrets.GOOGLE_CLOUD_PROJECT }}
           GOOGLE_CLOUD_LOCATION: ${{ secrets.GOOGLE_CLOUD_LOCATION }}
-          GOOGLE_GENAI_USE_VERTEXAI: 1
+          GOOGLE_GENAI_USE_ENTERPRISE: 1
           DOC_OWNER: 'google'
           DOC_REPO: 'adk-docs'
           CODE_OWNER: 'google'

.gitignore

@@ -199,3 +199,7 @@ docs/site
 # Ignore a local build but keep the directory
 /site/*
 !/site/.gitkeep
+
+# Compiled Go example binaries (accidentally committed build artifacts)
+/examples/go/index
+/examples/go/main

docs/2.0/index.md

@@ -5,10 +5,14 @@ hide:
 
 # Welcome to ADK 2.0
 
+<div class="language-support-tag">
+  <span class="lst-supported">Supported in ADK</span><span class="lst-python">Python v2.0.0</span><span class="lst-go">Go v2.0.0</span>
+</div>
+
 ADK 2.0 introduces powerful tools for building sophisticated AI agents, and
 helps you structure agents to execute challenging tasks with more control,
-predictability, and reliability. ADK 2.0 is available for Python and includes
-the following key features:
+predictability, and reliability. ADK 2.0 is available for Python and Go and
+includes the following key features:
 
 -   [**Graph-based workflows**](/graphs/): Build deterministic agent
     workflows with more control over how tasks are routed and executed.
@@ -28,6 +32,10 @@ to build agents with ADK 2.0!
 
     ADK Python 2.0 is released for general availability as of May 19, 2026.
 
+!!! tip "ADK Go v2.0.0 GA release"
+
+    ADK Go 2.0 is released for general availability as of June 30, 2026.
+
 ## ADK Python 1.x compatibility
 
 ADK 2.0 is designed to be compatible with agents developed with ADK 1.x
@@ -47,7 +55,7 @@ architecture, your Agents, Tools, and Functions are evaluated as individual
 following breaking changes and migration steps to ensure a smooth transition for
 your production applications.
 
-### Event Schema & Custom Session Databases
+### Event Schema & Custom Session Storage
 
 ADK 2.0 introduces new fields `node_info` and `output` to the core
 ***Event*** schema to track graph state and workflow outputs.
@@ -123,8 +131,8 @@ so the framework can evaluate them against your configured ***RetryConfig***,
 such as `RetryConfig(max_attempts=3)`. Never catch ***BaseException*** unless
 you are explicitly re-raising the exception.
 
-If you encounter additional ADK 1.0 to ADK 2.0 incompatibilities, report them
-through the
+If you encounter additional ADK Python 1.0 to ADK 2.0 incompatibilities, report
+them through the
 [issue tracker](https://github.com/google/adk-python/issues/new?template=bug_report.md&labels=v2).
 
 
@@ -176,6 +184,119 @@ To install the latest version of ADK 1.x, follow these steps:
         source .venv/bin/activate
         ```
 
+## ADK Go 1.x compatibility
+
+ADK Go 2.0 is designed to be compatible with agents developed with ADK Go 1.x
+releases. However, there are a few breaking changes you should be aware of
+before upgrading an ADK Go 1.x project to ADK Go 2.0.
+
+!!! warning "Breaking changes: ADK Go 1.x to 2.0 incompatibilities"
+
+    There are several known incompatibilities and breaking changes introduced
+    with ADK Go v2.0.0. Before upgrading, review these changes and take
+    mitigation steps, if necessary.
+
+The ADK Go 2.0 release introduces the Workflow Runtime, transitioning ADK Go
+from a hierarchical agent executor to a graph-based execution engine. In this
+new architecture, your Agents, Tools, and Functions are evaluated as individual
+*nodes* within a workflow graph. If you are upgrading from ADK Go 1.x, review
+the following breaking changes and migration steps.
+
+### Module import path
+
+ADK Go 2.0 uses a new major version module path. You must update all import
+paths in your Go source files and your `go.mod` file.
+
+*   **1.x import path:** `google.golang.org/adk`
+*   **2.0 import path:** `google.golang.org/adk/v2`
+
+**Migration action:** Run `go get google.golang.org/adk/v2` and update
+all import statements in your source files from `google.golang.org/adk/...` to
+`google.golang.org/adk/v2/...`.
+
+### Agent Execution: Agent interface changes
+
+In ADK Go 1.x, agents implemented the `agent.Agent` interface by providing a
+`Run` method. In ADK Go 2.0, agents are evaluated as individual *nodes* within
+the new Workflow Graph engine.
+
+*   **Execution driver custom overrides:** Custom agent types that override
+    internal execution behavior may no longer work as expected. The Workflow
+    Graph engine manages execution scheduling and event emission, and custom
+    implementations that bypass these mechanisms are silently ignored.
+
+**Migration action:** Move custom execution logic into standardized
+`BeforeAgentCallback` and `AfterAgentCallback` hooks to safely inject custom
+logic into the execution lifecycle.
+
+### Event Construction: `session.NewEvent` signature change
+
+`session.NewEvent` now requires a `context.Context` as its first argument:
+
+```go
+// Before (ADK Go 1.x)
+ev := session.NewEvent(ctx.InvocationID())
+// or
+ev := session.NewEventWithContext(ctx, ctx.InvocationID())
+
+// After (ADK Go 2.0)
+ev := session.NewEvent(ctx, ctx.InvocationID())
+```
+
+The event ID and timestamp are now obtained through the `platform` package,
+so a time or UUID provider installed on `ctx` controls them. This lets workflow
+engines produce deterministic, replay-safe events. The previous
+parameterless-context form and the temporary `NewEventWithContext` helper are
+removed.
+
+**Migration action:** Pass the context already in scope as the first argument
+to `session.NewEvent`. Any `context.Context` works — the `ctx` of an agent,
+tool, or callback (which embed `context.Context`), a request context, or in
+tests, `t.Context()`. If a helper that calls `NewEvent` does not yet receive a
+context, add a `ctx context.Context` parameter and thread it down from the
+caller. Avoid creating a new `context.Background()` mid-call-chain; reserve
+that for `main`, `init`, and top-level test setup.
+
+### Event Schema & Custom Session Storage
+
+ADK Go 2.0 adds five new fields to the core ***Event*** struct to support
+graph routing, workflow state, and human-in-the-loop pausing:
+
+| Go field | Serialized name | Purpose |
+|---|---|---|
+| `IsolationScope string` | `isolationScope` (`json:"isolationScope,omitempty"`) | Restricts which agent contexts see this event in LLM prompt history. |
+| `Routes []string` | `Routes` (no JSON tag) | Routing keys emitted by a node to drive conditional edge dispatch. |
+| `RequestedInput *RequestInput` | `RequestedInput` (no JSON tag) | Signals that a workflow node is pausing for human input. |
+| `Output any` | `Output` (no JSON tag) | Generic data output from a workflow node. |
+| `NodeInfo *NodeInfo` | `nodeInfo` (`json:"nodeInfo,omitempty"`) | Workflow-node metadata identifying which node emitted the event. |
+
+*   **Custom session storage:** If you have implemented a custom
+    `session.Service`, such as storing sessions in your own SQL or NoSQL
+    databases with rigid schemas, your underlying database schema must be
+    updated to accommodate all five new fields. Inserting a 2.0 ***Event***
+    into a rigid 1.x database table causes insertion or deserialization
+    failures. *However, if your custom session service stores events as
+    serialized JSON blobs, you do not need to update your schema.*
+
+**Migration action:** Update your database schemas and downstream client
+validators to expect and store the five new fields on all Event payloads.
+Pay particular attention to `Routes`, `RequestedInput`, and `Output`, which
+have no JSON struct tags and therefore serialize under their Go field names
+exactly as shown above.
+
+If you encounter additional ADK Go 1.0 to ADK 2.0 incompatibilities, report
+them through the
+[issue tracker](https://github.com/google/adk-go/issues/new?template=bug_report.md&labels=v2).
+
+### Installing ADK Go 1.x {#install-go}
+
+If you want to continue using ADK Go 1.x and are not yet ready to upgrade to
+ADK Go 2.0, pin your dependency to the 1.x release line:
+
+```shell
+go get google.golang.org/adk@v1
+```
+
 ## Next steps
 
 Read the developer guides for building agents with ADK 2.0 features:
@@ -186,8 +307,14 @@ Read the developer guides for building agents with ADK 2.0 features:
 
 Check out these ADK 2.0 code samples for testing and inspiration:
 
--   [**Workflow samples**](https://github.com/google/adk-python/tree/v2/contributing/workflow_samples)
--   [**Collaborative task samples**](https://github.com/google/adk-python/tree/v2/contributing/task_samples)
+=== "Python"
+
+    -   [**Workflow samples**](https://github.com/google/adk-python/tree/main/contributing/samples/workflows)
+    -   [**Collaborative task samples**](https://github.com/google/adk-python/tree/main/contributing/samples/multi_agent)
+
+=== "Go"
+
+    -   [**All workflow agents samples**](https://github.com/google/adk-go/tree/main/examples/workflow)
+    -   [**Collaborative task sample**](https://github.com/google/adk-go/tree/main/examples/multiagent/collaboration)
 
-Thanks for checking out ADK 2.0! We look forward to your
-[feedback](https://github.com/google/adk-python/issues/new?template=feature_request.md&labels=v2)!
+Thanks for checking out ADK 2.0! We look forward to your feedback — let us know on [ADK Go](https://github.com/google/adk-go/issues/new) or [ADK Python](https://github.com/google/adk-python/issues/new).

docs/agents/config.md

@@ -88,15 +88,15 @@ To create an ADK project for use with Agent Config:
     1.  For Gemini model access through Google API, add a line to the
         file with your API key:
 
-            GOOGLE_GENAI_USE_VERTEXAI=0
+            GOOGLE_GENAI_USE_ENTERPRISE=0
             GOOGLE_API_KEY=<your-Google-Gemini-API-key>
 
         You can get an API key from the Google AI Studio
         [API Keys](https://aistudio.google.com/app/apikey) page.
 
     1.  For Gemini model access through Google Cloud, add these lines to the file:
 
-            GOOGLE_GENAI_USE_VERTEXAI=1
+            GOOGLE_GENAI_USE_ENTERPRISE=1
             GOOGLE_CLOUD_PROJECT=<your_gcp_project>
             GOOGLE_CLOUD_LOCATION=us-central1
 

docs/agents/llm-agents.md

@@ -544,6 +544,28 @@ Control whether the agent receives the prior conversation history.
             .build();
     ```
 
+!!! note "Go v2.0.0: agent execution modes"
+
+    ADK Go v2.0.0 introduces an explicit `Mode` field on `llmagent.Config` that
+    controls how the agent runs when used inside a graph-based or dynamic
+    workflow. Three modes are available:
+
+    -   **`ModeChat`** (default for an agent used as a sub-agent): The agent
+        participates in a multi-turn conversation with the user and is reachable
+        from peer agents via `transfer_to_agent`.
+    -   **`ModeSingleTurn`** (default for an agent used as a node in a
+        workflow): The agent completes its task in a single turn without
+        chatting with the user.
+    -   **`ModeTask`**: A task agent that chats with the user to accomplish a
+        task — in contrast to `ModeSingleTurn`, it can interact with the user
+        across turns to complete the work.
+
+    When you wrap an `llmagent` with `workflow.NewAgentNode`, the workflow
+    engine automatically sets the mode to `ModeSingleTurn` if no mode is
+    specified — equivalent to Python's `mode="single_turn"` on an agent used
+    as a workflow node. For more information on composing agents in graph-based
+    workflows, see [Graph-based agent workflows](/graphs/).
+
 ### Planner
 
 <div class="language-support-tag" title="">
@@ -795,3 +817,6 @@ the following:
   planning (`planner`), controlling agent transfer
   (`disallow_transfer_to_parent`, `disallow_transfer_to_peers`), and system-wide
   instructions (`global_instruction`). See [Custom agent workflows](/agents/custom-agents/).
+* **Graph-based workflows:** Compose LLM agents as steps in deterministic,
+  graph-based pipelines using [Graph-based agent workflows](/graphs/). In Go v2.0.0, use
+  `workflow.NewAgentNode` to wrap any LLM agent as a workflow node.

docs/agents/models/agent-platform.md

@@ -147,7 +147,7 @@ Agent Platform.
     **Setup:**
 
     1. **Agent Platform Environment:** Ensure the consolidated Agent Platform setup (ADC, Env
-       Vars, `GOOGLE_GENAI_USE_VERTEXAI=TRUE`) is complete.
+       Vars, `GOOGLE_GENAI_USE_ENTERPRISE=TRUE`) is complete.
 
     2. **Install Provider Library:** Install the necessary client library configured
        for Agent Platform.
@@ -278,7 +278,7 @@ Agent Platform offers a curated selection of open-source models, such as Meta Ll
     **Setup:**
 
     1. **Agent Platform Environment:** Ensure the consolidated Agent Platform setup (ADC, Env
-       Vars, `GOOGLE_GENAI_USE_VERTEXAI=TRUE`) is complete.
+       Vars, `GOOGLE_GENAI_USE_ENTERPRISE=TRUE`) is complete.
 
     2. **Install LiteLLM:**
             ```shell

docs/agents/workflow-agents/index.md

@@ -12,11 +12,11 @@ how and when other agents run, defining the control flow of a process.
 
 !!! note "Alternative: graph-based workflows"
 
-    Starting in ADK 2.0, template workflows have been superseded
+    Starting in ADK 2.0 for Python and Go, template workflows have been superseded
 
     by more flexible workflow structures, including
-    [graph-based workflows](/workflows/graphs/) and
-    [dynamic workflows](/workflows/dynamic/).
+    [graph-based workflows](/graphs/) and
+    [dynamic workflows](/graphs/dynamic/).
     These workflow architectures provide more control, flexibility
     and capability to evolve your agent workflows over time.
 

docs/agents/workflow-agents/loop-agents.md

@@ -16,11 +16,11 @@ the ***LoopAgent*** object you define.
 
 !!! note "Alternative: graph-based workflows"
 
-    Starting in ADK 2.0, templated workflows have been superseded
+    Starting in ADK 2.0 for Python and Go, templated workflows have been superseded
 
     by more flexible workflow structures, including
-    [graph-based workflows](/workflows/graphs/) and
-    [dynamic workflows](/workflows/dynamic/).
+    [graph-based workflows](/graphs/) and
+    [dynamic workflows](/graphs/dynamic/).
 
 ### Example scenario
 

docs/agents/workflow-agents/parallel-agents.md

@@ -23,11 +23,11 @@ ultimately managed by the ***ParallelAgent*** object you define.
 
 !!! note "Alternative: graph-based workflows"
 
-    Starting in ADK 2.0, templated workflows have been superseded
+    Starting in ADK 2.0 for Python and Go, templated workflows have been superseded
 
     by more flexible workflow structures, including
-    [graph-based workflows](/workflows/graphs/) and
-    [dynamic workflows](/workflows/dynamic/).
+    [graph-based workflows](/graphs/) and
+    [dynamic workflows](/graphs/dynamic/).
 
 ### How it works
 

docs/agents/workflow-agents/sequential-agents.md

@@ -16,11 +16,11 @@ object you define.
 
 !!! note "Alternative: graph-based workflows"
 
-    Starting in ADK 2.0, templated workflows have been superseded
+    Starting in ADK 2.0 for Python and Go, templated workflows have been superseded
 
     by more flexible workflow structures, including
-    [graph-based workflows](/workflows/graphs/) and
-    [dynamic workflows](/workflows/dynamic/).
+    [graph-based workflows](/graphs/) and
+    [dynamic workflows](/graphs/dynamic/).
 
 ### Example scenario
 

docs/deploy/cloud-run.md

@@ -56,7 +56,7 @@ Set your environment variables as described in the [Setup and Installation](../g
 ```bash
 export GOOGLE_CLOUD_PROJECT=your-project-id
 export GOOGLE_CLOUD_LOCATION=us-central1 # Or your preferred location
-export GOOGLE_GENAI_USE_VERTEXAI=True
+export GOOGLE_GENAI_USE_ENTERPRISE=True
 ```
 
 For more information on connecting to Google Cloud from ADK agents, see
@@ -327,7 +327,7 @@ unless you specify it as deployment setting, such as the `--with_ui` option for
     --region $GOOGLE_CLOUD_LOCATION \
     --project $GOOGLE_CLOUD_PROJECT \
     --allow-unauthenticated \
-    --set-env-vars="GOOGLE_CLOUD_PROJECT=$GOOGLE_CLOUD_PROJECT,GOOGLE_CLOUD_LOCATION=$GOOGLE_CLOUD_LOCATION,GOOGLE_GENAI_USE_VERTEXAI=$GOOGLE_GENAI_USE_VERTEXAI"
+    --set-env-vars="GOOGLE_CLOUD_PROJECT=$GOOGLE_CLOUD_PROJECT,GOOGLE_CLOUD_LOCATION=$GOOGLE_CLOUD_LOCATION,GOOGLE_GENAI_USE_ENTERPRISE=$GOOGLE_GENAI_USE_ENTERPRISE"
     # Add any other necessary environment variables your agent might need
     ```
 
@@ -570,7 +570,7 @@ unless you specify it as deployment setting, such as the `--with_ui` option for
     --region $GOOGLE_CLOUD_LOCATION \
     --project $GOOGLE_CLOUD_PROJECT \
     --allow-unauthenticated \
-    --set-env-vars="GOOGLE_CLOUD_PROJECT=$GOOGLE_CLOUD_PROJECT,GOOGLE_CLOUD_LOCATION=$GOOGLE_CLOUD_LOCATION,GOOGLE_GENAI_USE_VERTEXAI=$GOOGLE_GENAI_USE_VERTEXAI"
+    --set-env-vars="GOOGLE_CLOUD_PROJECT=$GOOGLE_CLOUD_PROJECT,GOOGLE_CLOUD_LOCATION=$GOOGLE_CLOUD_LOCATION,GOOGLE_GENAI_USE_ENTERPRISE=$GOOGLE_GENAI_USE_ENTERPRISE"
     # Add any other necessary environment variables your agent might need
     ```
 

docs/deploy/gke.md

@@ -8,7 +8,7 @@
 
 To deploy your agent you will need to have a Kubernetes cluster running on GKE. You can create a cluster using the Google Cloud Console or the `gcloud` command line tool.
 
-In this example we will deploy a simple agent to GKE. The Python agent is a FastAPI application that uses `Gemini Flash` as the LLM. The Go agent uses the ADK launcher and a statically-linked binary in a minimal container. We can use Vertex AI or AI Studio as the LLM provider.
+In this example deploys a simple agent to GKE. The Python agent is a FastAPI application that uses `Gemini Flash` as the LLM. The Go agent uses the ADK launcher and a statically-linked binary in a minimal container. We can use Vertex AI or AI Studio as the LLM provider. You can use Agent Platform or AI Studio as the LLM provider with the environment variable `GOOGLE_GENAI_USE_ENTERPRISE`.
 
 ## Environment variables
 
@@ -17,7 +17,7 @@ Set your environment variables as described in the [Setup and Installation](../g
 ```bash
 export GOOGLE_CLOUD_PROJECT=your-project-id # Your GCP project ID
 export GOOGLE_CLOUD_LOCATION=us-central1 # Or your preferred location
-export GOOGLE_GENAI_USE_VERTEXAI=true # Set to true if using Agent Platform
+export GOOGLE_GENAI_USE_ENTERPRISE=true # Set to true if using Agent Platform
 export GOOGLE_CLOUD_PROJECT_NUMBER=$(gcloud projects describe --format json $GOOGLE_CLOUD_PROJECT | jq -r ".projectNumber")
 ```
 
@@ -479,9 +479,9 @@ spec:
             value: $GOOGLE_CLOUD_PROJECT
           - name: GOOGLE_CLOUD_LOCATION
             value: $GOOGLE_CLOUD_LOCATION
-          - name: GOOGLE_GENAI_USE_VERTEXAI
-            value: "$GOOGLE_GENAI_USE_VERTEXAI"
-          # If using AI Studio, set GOOGLE_GENAI_USE_VERTEXAI to false and set the following:
+          - name: GOOGLE_GENAI_USE_ENTERPRISE
+            value: "$GOOGLE_GENAI_USE_ENTERPRISE"
+          # If using AI Studio, set GOOGLE_GENAI_USE_ENTERPRISE to false and set the following:
           # - name: GOOGLE_API_KEY
           #   value: $GOOGLE_API_KEY
           # Add any other necessary environment variables your agent might need