chore: Bumps version to v1.21.0 and updates CHANGELOG.md

Co-authored-by: Xiang (Sean) Zhou <seanzhougoogle@google.com>
PiperOrigin-RevId: 843427582

.gitignore

@@ -101,6 +101,7 @@ Thumbs.db
 
 # AI Coding Tools - Project-specific configs
 # Developers should symlink or copy AGENTS.md and add their own overrides locally
+.adk/
 .claude/
 CLAUDE.md
 .cursor/

CHANGELOG.md

@@ -1,5 +1,75 @@
 # Changelog
 
+## [1.21.0](https://github.com/google/adk-python/compare/v1.20.0...v1.21.0) (2025-12-11)
+
+### Features
+* **[Interactions API Support]**
+  * The newly released Gemini [Interactions API](https://ai.google.dev/gemini-api/docs/interactions) is supported in ADK Now. To use it:
+  ```Python
+  Agent(
+    model=Gemini(
+        model="gemini-3-pro-preview",
+        use_interactions_api=True,
+    ),
+    name="...",
+    description="...",
+    instruction="...",
+  )
+  ```
+  see [samples](https://github.com/google/adk-python/tree/main/contributing/samples/interactions_api) for details
+
+
+* **[Services]**
+  * Add `add_session_to_memory` to `CallbackContext` and `ToolContext` to explicitly save the current session to memory ([7b356dd](https://github.com/google/adk-python/commit/7b356ddc1b1694d2c8a9eee538f3a41cf5518e42))
+
+* **[Plugins]**
+  * Add location for table in agent events in plugin BigQueryAgentAnalytics ([507424a](https://github.com/google/adk-python/commit/507424acb9aabc697fc64ef2e9a57875f25f0a21))
+  * Upgrade BigQueryAgentAnalyticsPlugin to v2.0 with improved performance, multimodal support, and reliability ([7b2fe14](https://github.com/google/adk-python/commit/7b2fe14dab96440ee25b66dae9e66eadba629a56))
+
+
+* **[A2A]**
+  * Adds ADK EventActions to A2A response ([32e87f6](https://github.com/google/adk-python/commit/32e87f6381ff8905a06a9a43a0207d758a74299d))
+
+* **[Tools]**
+  * Add `header_provider` to `OpenAPIToolset` and `RestApiTool` ([e1a7593](https://github.com/google/adk-python/commit/e1a7593ae8455d51cdde46f5165410217400d3c9))
+  * Allow overriding connection template ([cde7f7c](https://github.com/google/adk-python/commit/cde7f7c243a7cdc8c7b886f68be55fd59b1f6d5a))
+  * Add SSL certificate verification configuration to OpenAPI tools using the `verify` parameter ([9d2388a](https://github.com/google/adk-python/commit/9d2388a46f7a481ea1ec522f33641a06c64394ed))
+  * Use json schema for function tool declaration when feature enabled ([cb3244b](https://github.com/google/adk-python/commit/cb3244bb58904ab508f77069b436f85b442d3299))
+
+* **[Models]**
+  * Add Gemma3Ollama model integration and a sample ([e9182e5](https://github.com/google/adk-python/commit/e9182e5eb4a37fb5219fc607cd8f06d7e6982e83))
+
+
+### Bug Fixes
+
+* Install dependencies for py 3.10 ([9cccab4](https://github.com/google/adk-python/commit/9cccab453706138826f313c47118812133e099c4))
+* Refactor LiteLLM response schema formatting for different models ([894d8c6](https://github.com/google/adk-python/commit/894d8c6c2652492324c428e8dae68a8646b17485))
+* Resolve project and credentials before creating Spanner client ([99f893a](https://github.com/google/adk-python/commit/99f893ae282a04c67cce5f80e87d3bfadd3943e6))
+* Avoid false positive "App name mismatch" warnings in Runner ([6388ba3](https://github.com/google/adk-python/commit/6388ba3b2054e60d218eae6ec8abc621ed0a1139))
+* Update the code to work with either 1 event or more than 1 events ([4f54660](https://github.com/google/adk-python/commit/4f54660d6de54ddde0fec6e09fdd68890ce657ca))
+* OpenAPI schema generation by skipping JSON schema for judge_model_config ([56775af](https://github.com/google/adk-python/commit/56775afc48ee54e9cbea441a6e0fa6c8a12891b9))
+* Add tool_name_prefix support to OpenAPIToolset ([82e6623](https://github.com/google/adk-python/commit/82e6623fa97fb9cbc6893b44e228f4da098498da))
+* Pass context to client interceptors ([143ad44](https://github.com/google/adk-python/commit/143ad44f8c5d1c56fc92dd691589aaa0b788e485))
+* Yield event with error code when agent run raised A2AClientHTTPError ([b7ce5e1](https://github.com/google/adk-python/commit/b7ce5e17b6653074c5b41d08b2027b5e9970a671))
+* Handle string function responses in LiteLLM conversion ([2b64715](https://github.com/google/adk-python/commit/2b6471550591ee7fc5f70f79e66a6e4080df442b))
+* ApigeeLLM support for Built-in tools like GoogleSearch, BuiltInCodeExecutor when calling Gemini models through Apigee ([a9b853f](https://github.com/google/adk-python/commit/a9b853fe364d08703b37914a89cf02293b5c553b))
+* Extract and propagate task_id in RemoteA2aAgent ([82bd4f3](https://github.com/google/adk-python/commit/82bd4f380bd8b4822191ea16e6140fe2613023ad))
+* Update FastAPI and Starlette to fix CVE-2025-62727 (ReDoS vulnerability) ([c557b0a](https://github.com/google/adk-python/commit/c557b0a1f2aac9f0ef7f1e0f65e3884007407e30))
+* Add client id to token exchange ([f273517](https://github.com/google/adk-python/commit/f2735177f195b8d7745dba6360688ddfebfed31a))
+
+### Improvements
+
+* Normalize multipart content for LiteLLM's ollama_chat provider ([055dfc7](https://github.com/google/adk-python/commit/055dfc79747aa365db8441908d4994f795e94a68))
+* Update adk web, fixes image not rendering, state not updating, update drop down box width and trace icons ([df86847](https://github.com/google/adk-python/commit/df8684734bbfd5a8afe3b4362574fe93dcb43048))
+* Add sample agent for interaction api integration ([68d7048](https://github.com/google/adk-python/commit/68d70488b9340251a9d37e8ae3a9166870f26aa1))
+* Update genAI SDK version ([f0bdcab](https://github.com/google/adk-python/commit/f0bdcaba449f21bd8c27cde7dbedc03bf5ec5349))
+* Introduce `build_function_declaration_with_json_schema` to use pydantic to generate json schema for FunctionTool ([51a638b](https://github.com/google/adk-python/commit/51a638b6b85943d4aaec4ee37c95a55386ebac90))
+* Update component definition for triaging agent ([ee743bd](https://github.com/google/adk-python/commit/ee743bd19a8134129111fc4769ec24e40a611982))
+* Migrate Google tools to use the new feature decorator ([bab5729](https://github.com/google/adk-python/commit/bab57296d553cb211106ece9ee2c226c64a60c57))
+* Migrate computer to use the new feature decorator ([1ae944b](https://github.com/google/adk-python/commit/1ae944b39d9cf263e15b36c76480975fe4291d22))
+* Add Spanner execute sql query result mode using list of dictionaries ([f22bac0](https://github.com/google/adk-python/commit/f22bac0b202cd8f273bf2dee9fff57be1b40730d))
+* Improve error message for missing `invocation_id` and `new_message` in `run_async` ([de841a4](https://github.com/google/adk-python/commit/de841a4a0982d98ade4478f10481c817a923faa2))
+
 ## [1.20.0](https://github.com/google/adk-python/compare/v1.19.0...v1.20.0) (2025-12-01)
 
 

contributing/samples/adk_documentation/adk_docs_updater/main.py

@@ -24,13 +24,14 @@ from adk_documentation.settings import DOC_OWNER
 from adk_documentation.settings import DOC_REPO
 from adk_documentation.tools import get_issue
 from adk_documentation.utils import call_agent_async
+from adk_documentation.utils import parse_suggestions
 from google.adk.cli.utils import logs
 from google.adk.runners import InMemoryRunner
 
 APP_NAME = "adk_docs_updater"
 USER_ID = "adk_docs_updater_user"
 
-logs.setup_adk_logger(level=logging.DEBUG)
+logs.setup_adk_logger(level=logging.INFO)
 
 
 def process_arguments():
@@ -68,23 +69,84 @@ async def main():
     print(f"Failed to get issue {issue_number}: {get_issue_response}\n")
     return
   issue = get_issue_response["issue"]
+  issue_title = issue.get("title", "")
+  issue_body = issue.get("body", "")
+
+  # Parse numbered suggestions from issue body
+  suggestions = parse_suggestions(issue_body)
+
+  if not suggestions:
+    print(f"No numbered suggestions found in issue #{issue_number}.")
+    print("Falling back to processing the entire issue as a single task.")
+    suggestions = [(1, issue_body)]
+
+  print(f"Found {len(suggestions)} suggestion(s) in issue #{issue_number}.")
+  print("=" * 80)
 
   runner = InMemoryRunner(
       agent=agent.root_agent,
       app_name=APP_NAME,
   )
-  session = await runner.session_service.create_session(
-      app_name=APP_NAME,
-      user_id=USER_ID,
-  )
 
-  response = await call_agent_async(
-      runner,
-      USER_ID,
-      session.id,
-      f"Please update the ADK docs according to the following issue:\n{issue}",
+  results = []
+  for suggestion_num, suggestion_text in suggestions:
+    print(f"\n>>> Processing suggestion #{suggestion_num}...")
+    print("-" * 80)
+
+    # Create a new session for each suggestion to avoid context interference
+    session = await runner.session_service.create_session(
+        app_name=APP_NAME,
+        user_id=USER_ID,
+    )
+
+    prompt = f"""
+      Please update the ADK docs according to suggestion #{suggestion_num} from issue #{issue_number}.
+
+      Issue title: {issue_title}
+
+      Suggestion to process:
+      {suggestion_text}
+
+      Note: Focus only on this specific suggestion. Create exactly one pull request for this suggestion.
+    """
+
+    try:
+      response = await call_agent_async(
+          runner,
+          USER_ID,
+          session.id,
+          prompt,
+      )
+      results.append({
+          "suggestion_num": suggestion_num,
+          "status": "success",
+          "response": response,
+      })
+      print(f"<<<< Suggestion #{suggestion_num} completed.")
+    except Exception as e:
+      results.append({
+          "suggestion_num": suggestion_num,
+          "status": "error",
+          "error": str(e),
+      })
+      print(f"<<<< Suggestion #{suggestion_num} failed: {e}")
+
+    print("-" * 80)
+
+  # Print summary
+  print("\n" + "=" * 80)
+  print("SUMMARY")
+  print("=" * 80)
+  successful = [r for r in results if r["status"] == "success"]
+  failed = [r for r in results if r["status"] == "error"]
+  print(
+      f"Total: {len(results)}, Success: {len(successful)}, Failed:"
+      f" {len(failed)}"
   )
-  print(f"<<<< Agent Final Output: {response}\n")
+  if failed:
+    print("\nFailed suggestions:")
+    for r in failed:
+      print(f"  - Suggestion #{r['suggestion_num']}: {r['error']}")
 
 
 if __name__ == "__main__":

contributing/samples/adk_documentation/utils.py

@@ -12,9 +12,11 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
+import re
 from typing import Any
 from typing import Dict
 from typing import List
+from typing import Tuple
 
 from adk_documentation.settings import GITHUB_TOKEN
 from google.adk.agents.run_config import RunConfig
@@ -96,3 +98,47 @@ async def call_agent_async(
           final_response_text += text
 
   return final_response_text
+
+
+def parse_suggestions(issue_body: str) -> List[Tuple[int, str]]:
+  """Parse numbered suggestions from issue body.
+
+  Supports multiple formats:
+  - Format A (markdown headers): "### 1. Title"
+  - Format B (numbered list with bold): "1. **Title**"
+
+  Args:
+      issue_body: The body text of the GitHub issue.
+
+  Returns:
+      A list of tuples, where each tuple contains:
+      - The suggestion number (1-based)
+      - The full text of that suggestion
+  """
+  # Try different patterns in order of preference
+  patterns = [
+      # Format A: "### 1. Title" (markdown header with number)
+      (r"(?=^###\s+\d+\.)", r"^###\s+(\d+)\."),
+      # Format B: "1. **Title**" (numbered list with bold)
+      (r"(?=^\d+\.\s+\*\*)", r"^(\d+)\.\s+\*\*"),
+  ]
+
+  for split_pattern, match_pattern in patterns:
+    parts = re.split(split_pattern, issue_body, flags=re.MULTILINE)
+
+    suggestions = []
+    for part in parts:
+      part = part.strip()
+      if not part:
+        continue
+
+      match = re.match(match_pattern, part)
+      if match:
+        suggestion_num = int(match.group(1))
+        suggestions.append((suggestion_num, part))
+
+    # If we found suggestions with this pattern, return them
+    if suggestions:
+      return suggestions
+
+  return []

contributing/samples/adk_triaging_agent/README.md

@@ -18,7 +18,7 @@ The agent performs different actions based on the issue state:
 
 ### Component Labels
 The agent can assign the following component labels, each mapped to an owner:
-- `core`, `tools`, `mcp`, `eval`, `live`, `models`, `tracing`, `web`, `services`, `documentation`, `question`, `agent engine`, `a2a`, `bq`
+- `a2a`, `agent engine`, `auth`, `bq`, `core`, `documentation`, `eval`, `live`, `mcp`, `models`, `services`, `tools`, `tracing`, `web`, `workflow`
 
 ### Issue Types
 Based on the issue content, the agent will set the issue type to:

contributing/samples/adk_triaging_agent/agent.py

@@ -58,13 +58,16 @@ LABEL_GUIDELINES = """
       - "tracing": Telemetry, observability, structured logs, or spans.
       - "core": Core ADK runtime (Agent definitions, Runner, planners,
         thinking config, CLI commands, GlobalInstructionPlugin, CPU usage, or
-        general orchestration). Default to "core" when the topic is about ADK
-        behavior and no other label is a better fit.
+        general orchestration including agent transfer for multi-agents system).
+        Default to "core" when the topic is about ADK behavior and no other
+        label is a better fit.
       - "agent engine": Vertex AI Agent Engine deployment or sandbox topics
         only (e.g., `.agent_engine_config.json`, `ae_ignore`, Agent Engine
         sandbox, `agent_engine_id`). If the issue does not explicitly mention
         Agent Engine concepts, do not use this label—choose "core" instead.
-      - "a2a": Agent-to-agent workflows, coordination logic, or A2A protocol.
+      - "a2a": A2A protocol, running agent as a2a agent with "--a2a" option for
+        remote agent to talk with. Talking to remote agent via RemoteA2aAgent.
+        NOT including those local multi-agent systems.
       - "bq": BigQuery integration or general issues related to BigQuery.
       - "workflow": Workflow agents and workflow execution.
       - "auth": Authentication or authorization issues.
@@ -253,25 +256,6 @@ root_agent = Agent(
 
       {LABEL_GUIDELINES}
 
-      Here are the rules for labeling:
-      - If the user is asking about documentation-related questions, label it with "documentation".
-      - If it's about session, memory services, label it with "services".
-      - If it's about UI/web, label it with "web".
-      - If the user is asking about a question, label it with "question".
-      - If it's related to tools, label it with "tools".
-      - If it's about agent evaluation, then label it with "eval".
-      - If it's about streaming/live, label it with "live".
-      - If it's about model support (non-Gemini, like Litellm, Ollama, OpenAI models), label it with "models".
-      - If it's about tracing, label it with "tracing".
-      - If it's agent orchestration, agent definition, Runner behavior, planners, or performance, label it with "core".
-      - Use "agent engine" only when the issue clearly references Vertex AI Agent Engine deployment artifacts (for example `.agent_engine_config.json`, `ae_ignore`, `agent_engine_id`, or Agent Engine sandbox errors).
-      - If it's about Model Context Protocol (e.g. MCP tool, MCP toolset, MCP session management etc.), label it with both "mcp" and "tools".
-      - If it's about A2A integrations or workflows, label it with "a2a".
-      - If it's about BigQuery integrations, label it with "bq".
-      - If it's about workflow agents or workflow execution, label it with "workflow".
-      - If it's about authentication, label it with "auth".
-      - If you can't find an appropriate labels for the issue, follow the previous instruction that starts with "IMPORTANT:".
-
       ## Triaging Workflow
 
       Each issue will have flags indicating what actions are needed:

contributing/samples/hello_world_gemma3_ollama/__init__.py

@@ -0,0 +1,16 @@
+# Copyright 2025 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+
+from . import agent

contributing/samples/hello_world_gemma3_ollama/agent.py

@@ -0,0 +1,93 @@
+# Copyright 2025 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import logging
+import random
+
+from google.adk.agents.llm_agent import Agent
+from google.adk.models import Gemma3Ollama
+
+litellm_logger = logging.getLogger("LiteLLM")
+litellm_logger.setLevel(logging.WARNING)
+
+
+def roll_die(sides: int) -> int:
+  """Roll a die and return the rolled result.
+
+  Args:
+    sides: The integer number of sides the die has.
+
+  Returns:
+    An integer of the result of rolling the die.
+  """
+  return random.randint(1, sides)
+
+
+async def check_prime(nums: list[int]) -> str:
+  """Check if a given list of numbers are prime.
+
+  Args:
+    nums: The list of numbers to check.
+
+  Returns:
+    A str indicating which number is prime.
+  """
+  primes = set()
+  for number in nums:
+    number = int(number)
+    if number <= 1:
+      continue
+    is_prime = True
+    for i in range(2, int(number**0.5) + 1):
+      if number % i == 0:
+        is_prime = False
+        break
+    if is_prime:
+      primes.add(number)
+  return (
+      "No prime numbers found."
+      if not primes
+      else f"{', '.join(str(num) for num in primes)} are prime numbers."
+  )
+
+
+root_agent = Agent(
+    model=Gemma3Ollama(),
+    name="data_processing_agent",
+    description=(
+        "hello world agent that can roll a dice of 8 sides and check prime"
+        " numbers."
+    ),
+    instruction="""
+      You roll dice and answer questions about the outcome of the dice rolls.
+      You can roll dice of different sizes.
+      You can use multiple tools in parallel by calling functions in parallel (in one request and in one round).
+      It is ok to discuss previous dice rolls, and comment on the dice rolls.
+      When you are asked to roll a die, you must call the roll_die tool with the number of sides. Be sure to pass in an integer. Do not pass in a string.
+      You should never roll a die on your own.
+      When checking prime numbers, call the check_prime tool with a list of integers. Be sure to pass in a list of integers. You should never pass in a string.
+      You should not check prime numbers before calling the tool.
+      When you are asked to roll a die and check prime numbers, you should always make the following two function calls:
+      1. You should first call the roll_die tool to get a roll. Wait for the function response before calling the check_prime tool.
+      2. After you get the function response from roll_die tool, you should call the check_prime tool with the roll_die result.
+        2.1 If user asks you to check primes based on previous rolls, make sure you include the previous rolls in the list.
+      3. When you respond, you must include the roll_die result from step 1.
+      You should always perform the previous 3 steps when asking for a roll and checking prime numbers.
+      You should not rely on the previous history on prime results.
+    """,
+    tools=[
+        roll_die,
+        check_prime,
+    ],
+)

contributing/samples/hello_world_gemma3_ollama/main.py

@@ -0,0 +1,77 @@
+# Copyright 2025 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+
+import asyncio
+import time
+
+import agent
+from dotenv import load_dotenv
+from google.adk.artifacts.in_memory_artifact_service import InMemoryArtifactService
+from google.adk.cli.utils import logs
+from google.adk.runners import Runner
+from google.adk.sessions.in_memory_session_service import InMemorySessionService
+from google.adk.sessions.session import Session
+from google.genai import types
+
+load_dotenv(override=True)
+logs.log_to_tmp_folder()
+
+
+async def main():
+
+  app_name = 'my_app'
+  user_id_1 = 'user1'
+  session_service = InMemorySessionService()
+  artifact_service = InMemoryArtifactService()
+  runner = Runner(
+      app_name=app_name,
+      agent=agent.root_agent,
+      artifact_service=artifact_service,
+      session_service=session_service,
+  )
+  session_1 = await session_service.create_session(
+      app_name=app_name, user_id=user_id_1
+  )
+
+  async def run_prompt(session: Session, new_message: str):
+    content = types.Content(
+        role='user', parts=[types.Part.from_text(text=new_message)]
+    )
+    print('** User says:', content.model_dump(exclude_none=True))
+    async for event in runner.run_async(
+        user_id=user_id_1,
+        session_id=session.id,
+        new_message=content,
+    ):
+      if event.content.parts and event.content.parts[0].text:
+        print(f'** {event.author}: {event.content.parts[0].text}')
+
+  start_time = time.time()
+  print('Start time:', start_time)
+  print('------------------------------------')
+  await run_prompt(session_1, 'Hi, introduce yourself.')
+  await run_prompt(
+      session_1, 'Roll a die with 100 sides and check if it is prime'
+  )
+  await run_prompt(session_1, 'Roll it again.')
+  await run_prompt(session_1, 'What numbers did I get?')
+  end_time = time.time()
+  print('------------------------------------')
+  print('End time:', end_time)
+  print('Total time:', end_time - start_time)
+
+
+if __name__ == '__main__':
+  asyncio.run(main())

contributing/samples/human_in_loop/README.md

@@ -18,7 +18,7 @@ This example demonstrates an agent using a long-running tool (`ask_for_approval`
     # Example: After external approval
     updated_tool_output_data = {
         "status": "approved",
-        "ticket-id": ticket_id, # from original call
+        "ticketId": ticket_id, # from original call
         # ... other relevant updated data
     }
 
@@ -31,12 +31,13 @@ This example demonstrates an agent using a long-running tool (`ask_for_approval`
     )
 
     # Send this back to the agent
-    await runner.run_async(
+    async for _ in runner.run_async(
         # ... session_id, user_id ...
         new_message=types.Content(
             parts=[updated_function_response_part], role="user"
         ),
-    )
+    ):
+        pass  # exhaust generator (or handle events)
     ```
 6.  **Agent Acts on Update**: The agent receives this message containing the `types.FunctionResponse` and, based on its instructions, proceeds with the next steps (e.g., calling another tool like `reimburse`).
 

contributing/samples/interactions_api/README.md

@@ -0,0 +1,153 @@
+# Interactions API Sample Agent
+
+This sample agent demonstrates the Interactions API integration in ADK. The
+Interactions API provides stateful conversation capabilities, allowing chained
+interactions using `previous_interaction_id` instead of sending full
+conversation history.
+
+## Features Tested
+
+1. **Basic Text Generation** - Simple conversation without tools
+2. **Google Search Tool** - Web search using `GoogleSearchTool` with
+   `bypass_multi_tools_limit=True`
+3. **Multi-Turn Conversations** - Stateful interactions with context retention
+   via `previous_interaction_id`
+4. **Custom Function Tool** - Weather lookup using `get_current_weather`
+
+## Important: Tool Compatibility
+
+The Interactions API does **NOT** support mixing custom function calling tools
+with built-in tools (like `google_search`) in the same agent. To work around
+this limitation:
+
+```python
+# Use bypass_multi_tools_limit=True to convert google_search to a function tool
+GoogleSearchTool(bypass_multi_tools_limit=True)
+```
+
+This converts the built-in `google_search` to a function calling tool (via
+`GoogleSearchAgentTool`), which allows it to work alongside custom function
+tools.
+
+## How to Run
+
+### Prerequisites
+
+```bash
+# From the adk-python root directory
+uv sync --all-extras
+source .venv/bin/activate
+
+# Set up authentication (choose one):
+# Option 1: Using Google Cloud credentials
+export GOOGLE_CLOUD_PROJECT=your-project-id
+
+# Option 2: Using API Key
+export GOOGLE_API_KEY=your-api-key
+```
+
+### Running Tests
+
+```bash
+cd contributing/samples
+
+# Run automated tests with Interactions API
+python -m interactions_api.main
+```
+
+## Key Differences: Interactions API vs Standard API
+
+### Interactions API (`use_interactions_api=True`)
+- Uses stateful interactions via `previous_interaction_id`
+- Only sends current turn contents when chaining interactions
+- Returns `interaction_id` in responses for chaining
+- Ideal for long conversations with many turns
+- Context caching is not used (state maintained via interaction chaining)
+
+### Standard API (`use_interactions_api=False`)
+- Uses stateless `generate_content` calls
+- Sends full conversation history with each request
+- No interaction IDs in responses
+- Context caching can be used
+
+## Code Structure
+
+```
+interactions_api/
+├── __init__.py                    # Package initialization
+├── agent.py                       # Agent definition with Interactions API
+├── main.py                        # Test runner
+├── test_interactions_curl.sh      # cURL-based API tests
+├── test_interactions_direct.py    # Direct API tests
+└── README.md                      # This file
+```
+
+## Agent Configuration
+
+```python
+from google.adk.agents.llm_agent import Agent
+from google.adk.models.google_llm import Gemini
+from google.adk.tools.google_search_tool import GoogleSearchTool
+
+root_agent = Agent(
+    model=Gemini(
+        model="gemini-2.5-flash",
+        use_interactions_api=True,  # Enable Interactions API
+    ),
+    name="interactions_test_agent",
+    tools=[
+        GoogleSearchTool(bypass_multi_tools_limit=True),  # Converted to function tool
+        get_current_weather,  # Custom function tool
+    ],
+)
+```
+
+## Example Output
+
+```
+============================================================
+TEST 1: Basic Text Generation
+============================================================
+
+>> User: Hello! What can you help me with?
+<< Agent: Hello! I can help you with: 1) Search the web...
+   [Interaction ID: v1_abc123...]
+PASSED: Basic text generation works
+
+============================================================
+TEST 2: Function Calling (Google Search Tool)
+============================================================
+
+>> User: Search for the capital of France.
+   [Tool Call] google_search_agent({'request': 'capital of France'})
+   [Tool Result] google_search_agent: {'result': 'The capital of France is Paris...'}
+<< Agent: The capital of France is Paris.
+   [Interaction ID: v1_def456...]
+PASSED: Google search tool works
+
+============================================================
+TEST 3: Multi-Turn Conversation (Stateful)
+============================================================
+
+>> User: Remember the number 42.
+<< Agent: I'll remember that number - 42.
+   [Interaction ID: v1_ghi789...]
+
+>> User: What number did I ask you to remember?
+<< Agent: You asked me to remember the number 42.
+   [Interaction ID: v1_jkl012...]
+PASSED: Multi-turn conversation works with context retention
+
+============================================================
+TEST 5: Custom Function Tool (get_current_weather)
+============================================================
+
+>> User: What's the weather like in Tokyo?
+   [Tool Call] get_current_weather({'city': 'Tokyo'})
+   [Tool Result] get_current_weather: {'city': 'Tokyo', 'temperature_f': 68, ...}
+<< Agent: The weather in Tokyo is 68F and Partly Cloudy.
+   [Interaction ID: v1_mno345...]
+PASSED: Custom function tool works with bypass_multi_tools_limit
+
+ALL TESTS PASSED (Interactions API)
+```

contributing/samples/interactions_api/__init__.py

@@ -0,0 +1,17 @@
+# Copyright 2025 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+"""Sample agent for testing the Interactions API integration."""
+
+from . import agent

contributing/samples/interactions_api/agent.py

@@ -0,0 +1,105 @@
+# Copyright 2025 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+"""Agent definition for testing the Interactions API integration.
+
+NOTE: The Interactions API does NOT support mixing custom function calling tools
+with built-in tools in the same agent. To work around this limitation, we use
+bypass_multi_tools_limit=True on GoogleSearchTool, which converts the built-in
+google_search to a function calling tool (via GoogleSearchAgentTool).
+
+The bypass is only triggered when len(agent.tools) > 1, so we include multiple
+tools in the agent (GoogleSearchTool + get_current_weather).
+
+With bypass_multi_tools_limit=True and multiple tools, all tools become function
+calling tools, which allows mixing google_search with custom function tools.
+"""
+
+from google.adk.agents.llm_agent import Agent
+from google.adk.models.google_llm import Gemini
+from google.adk.tools.google_search_tool import GoogleSearchTool
+
+
+def get_current_weather(city: str) -> dict:
+  """Get the current weather for a city.
+
+  This is a mock implementation for testing purposes.
+
+  Args:
+    city: The name of the city to get weather for.
+
+  Returns:
+    A dictionary containing weather information.
+  """
+  # Mock weather data for testing
+  weather_data = {
+      "new york": {"temperature": 72, "condition": "Sunny", "humidity": 45},
+      "london": {"temperature": 59, "condition": "Cloudy", "humidity": 78},
+      "tokyo": {
+          "temperature": 68,
+          "condition": "Partly Cloudy",
+          "humidity": 60,
+      },
+      "paris": {"temperature": 64, "condition": "Rainy", "humidity": 85},
+      "sydney": {"temperature": 77, "condition": "Clear", "humidity": 55},
+  }
+
+  city_lower = city.lower()
+  if city_lower in weather_data:
+    data = weather_data[city_lower]
+    return {
+        "city": city,
+        "temperature_f": data["temperature"],
+        "condition": data["condition"],
+        "humidity": data["humidity"],
+    }
+  else:
+    return {
+        "city": city,
+        "temperature_f": 70,
+        "condition": "Unknown",
+        "humidity": 50,
+        "note": "Weather data not available, using defaults",
+    }
+
+
+# Main agent with google_search (via bypass) and custom function tools
+# Using bypass_multi_tools_limit=True converts google_search to a function calling tool.
+# We need len(tools) > 1 to trigger the bypass, so we include get_current_weather directly.
+# This allows mixing google_search with custom function tools via the Interactions API.
+#
+# NOTE: code_executor is not compatible with function calling mode because the model
+# tries to call a function (e.g., run_code) instead of outputting code in markdown.
+root_agent = Agent(
+    model=Gemini(
+        model="gemini-2.5-flash",
+        use_interactions_api=True,
+    ),
+    name="interactions_test_agent",
+    description="An agent for testing the Interactions API integration",
+    instruction="""You are a helpful assistant that can:
+
+1. Search the web for information using google_search
+2. Get weather information using get_current_weather
+
+When users ask for information that requires searching, use google_search.
+When users ask about weather, use get_current_weather.
+
+Be concise and helpful in your responses. Always confirm what you did.
+""",
+    tools=[
+        GoogleSearchTool(bypass_multi_tools_limit=True),
+        get_current_weather,
+    ],
+)

contributing/samples/interactions_api/main.py

@@ -0,0 +1,420 @@
+# Copyright 2025 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+"""Main script for testing the Interactions API integration.
+
+This script tests the following features:
+1. Basic text generation
+2. Google Search tool (via bypass_multi_tools_limit)
+3. Multi-turn conversations with stateful interactions
+4. Google Search tool (additional coverage)
+5. Custom function tool (get_current_weather)
+
+NOTE: The Interactions API does NOT support mixing custom function calling tools
+with built-in tools. To work around this, we use bypass_multi_tools_limit=True
+on GoogleSearchTool, which converts it to a function calling tool (via
+GoogleSearchAgentTool). The bypass only triggers when len(agent.tools) > 1,
+so we include both GoogleSearchTool and get_current_weather in the agent.
+
+NOTE: Code execution via UnsafeLocalCodeExecutor is not compatible with function
+calling mode because the model tries to call a function instead of outputting
+code in markdown.
+
+Run with:
+  cd contributing/samples
+  python -m interactions_api_test.main
+"""
+
+import argparse
+import asyncio
+import logging
+from pathlib import Path
+import time
+from typing import Optional
+
+from dotenv import load_dotenv
+from google.adk.agents.run_config import RunConfig
+from google.adk.cli.utils import logs
+from google.adk.runners import InMemoryRunner
+from google.adk.runners import Runner
+from google.genai import types
+
+from .agent import root_agent
+
+# Load .env from the samples directory (parent of this module's directory)
+_env_path = Path(__file__).parent.parent / ".env"
+load_dotenv(_env_path)
+
+APP_NAME = "interactions_api_test_app"
+USER_ID = "test_user"
+
+
+async def call_agent_async(
+    runner: Runner,
+    user_id: str,
+    session_id: str,
+    prompt: str,
+    agent_name: str = "",
+    show_interaction_id: bool = True,
+) -> tuple[str, Optional[str]]:
+  """Call the agent asynchronously with the user's prompt.
+
+  Args:
+    runner: The agent runner
+    user_id: The user ID
+    session_id: The session ID
+    prompt: The prompt to send
+    agent_name: The expected agent name for filtering responses
+    show_interaction_id: Whether to show interaction IDs in output
+
+  Returns:
+    A tuple of (response_text, interaction_id)
+  """
+  content = types.Content(
+      role="user", parts=[types.Part.from_text(text=prompt)]
+  )
+
+  final_response_text = ""
+  last_interaction_id = None
+
+  print(f"\n>> User: {prompt}")
+
+  async for event in runner.run_async(
+      user_id=user_id,
+      session_id=session_id,
+      new_message=content,
+      run_config=RunConfig(save_input_blobs_as_artifacts=False),
+  ):
+    # Track interaction ID if available
+    if event.interaction_id:
+      last_interaction_id = event.interaction_id
+
+    # Show function calls
+    if event.get_function_calls():
+      for fc in event.get_function_calls():
+        print(f"   [Tool Call] {fc.name}({fc.args})")
+
+    # Show function responses
+    if event.get_function_responses():
+      for fr in event.get_function_responses():
+        print(f"   [Tool Result] {fr.name}: {fr.response}")
+
+    # Collect text responses from the agent (not user, not partial)
+    if (
+        event.content
+        and event.content.parts
+        and event.author != "user"
+        and not event.partial
+    ):
+      for part in event.content.parts:
+        if part.text:
+          # Filter by agent name if provided, otherwise accept any non-user
+          if not agent_name or event.author == agent_name:
+            final_response_text += part.text
+
+  print(f"<< Agent: {final_response_text}")
+  if show_interaction_id and last_interaction_id:
+    print(f"   [Interaction ID: {last_interaction_id}]")
+
+  return final_response_text, last_interaction_id
+
+
+async def test_basic_text_generation(runner: Runner, session_id: str):
+  """Test basic text generation without tools."""
+  print("\n" + "=" * 60)
+  print("TEST 1: Basic Text Generation")
+  print("=" * 60)
+
+  response, interaction_id = await call_agent_async(
+      runner, USER_ID, session_id, "Hello! What can you help me with?"
+  )
+
+  assert response, "Expected a non-empty response"
+  print("PASSED: Basic text generation works")
+  return interaction_id
+
+
+async def test_function_calling(runner: Runner, session_id: str):
+  """Test function calling with the google_search tool."""
+  print("\n" + "=" * 60)
+  print("TEST 2: Function Calling (Google Search Tool)")
+  print("=" * 60)
+
+  response, interaction_id = await call_agent_async(
+      runner,
+      USER_ID,
+      session_id,
+      "Search for the capital of France.",
+  )
+
+  assert response, "Expected a non-empty response"
+  assert "paris" in response.lower(), f"Expected Paris in response: {response}"
+  print("PASSED: Google search tool works")
+  return interaction_id
+
+
+async def test_multi_turn_conversation(runner: Runner, session_id: str):
+  """Test multi-turn conversation to verify stateful interactions."""
+  print("\n" + "=" * 60)
+  print("TEST 3: Multi-Turn Conversation (Stateful)")
+  print("=" * 60)
+
+  # Turn 1: Tell the agent a fact directly (test conversation memory)
+  response1, id1 = await call_agent_async(
+      runner,
+      USER_ID,
+      session_id,
+      "My favorite color is blue. Just acknowledge this, don't use any tools.",
+  )
+  assert response1, "Expected a response for turn 1"
+  print(f"   Turn 1 interaction_id: {id1}")
+
+  # Turn 2: Ask about something else (use weather tool to add variety)
+  response2, id2 = await call_agent_async(
+      runner,
+      USER_ID,
+      session_id,
+      "What's the weather like in London?",
+  )
+  assert response2, "Expected a response for turn 2"
+  assert (
+      "59" in response2
+      or "london" in response2.lower()
+      or "cloudy" in response2.lower()
+  ), f"Expected London weather info in response: {response2}"
+  print(f"   Turn 2 interaction_id: {id2}")
+
+  # Turn 3: Ask the agent to recall conversation context
+  response3, id3 = await call_agent_async(
+      runner,
+      USER_ID,
+      session_id,
+      "What is my favorite color that I mentioned earlier in our conversation?",
+  )
+  assert response3, "Expected a response for turn 3"
+  assert (
+      "blue" in response3.lower()
+  ), f"Expected agent to remember the color 'blue': {response3}"
+  print(f"   Turn 3 interaction_id: {id3}")
+
+  # Verify interaction IDs are different (new interactions) but chained
+  if id1 and id2 and id3:
+    print(f"   Interaction chain: {id1} -> {id2} -> {id3}")
+
+  print("PASSED: Multi-turn conversation works with context retention")
+
+
+async def test_google_search_tool(runner: Runner, session_id: str):
+  """Test the google_search built-in tool."""
+  print("\n" + "=" * 60)
+  print("TEST 4: Google Search Tool (Additional)")
+  print("=" * 60)
+
+  response, interaction_id = await call_agent_async(
+      runner,
+      USER_ID,
+      session_id,
+      "Use google search to find out who wrote the novel '1984'.",
+  )
+
+  assert response, "Expected a non-empty response"
+  assert (
+      "orwell" in response.lower() or "george" in response.lower()
+  ), f"Expected George Orwell in response: {response}"
+  print("PASSED: Google search built-in tool works")
+
+
+async def test_custom_function_tool(runner: Runner, session_id: str):
+  """Test the custom function tool alongside google_search.
+
+  The root_agent has both GoogleSearchTool (with bypass_multi_tools_limit=True)
+  and get_current_weather. This tests that function calling tools work with
+  the Interactions API when all tools are function calling types.
+  """
+  print("\n" + "=" * 60)
+  print("TEST 5: Custom Function Tool (get_current_weather)")
+  print("=" * 60)
+
+  response, interaction_id = await call_agent_async(
+      runner,
+      USER_ID,
+      session_id,
+      "What's the weather like in Tokyo?",
+  )
+
+  assert response, "Expected a non-empty response"
+  # The mock weather data for Tokyo has temperature 68, condition "Partly Cloudy"
+  assert (
+      "68" in response
+      or "partly" in response.lower()
+      or "tokyo" in response.lower()
+  ), f"Expected weather info for Tokyo in response: {response}"
+  print("PASSED: Custom function tool works with bypass_multi_tools_limit")
+  return interaction_id
+
+
+def check_interactions_api_available() -> bool:
+  """Check if the interactions API is available in the SDK."""
+  try:
+    from google.genai import Client
+
+    client = Client()
+    # Check if interactions attribute exists
+    return hasattr(client.aio, "interactions")
+  except Exception:
+    return False
+
+
+async def run_all_tests():
+  """Run all tests with the Interactions API."""
+  print("\n" + "#" * 70)
+  print("# Running tests with Interactions API")
+  print("#" * 70)
+
+  # Check if interactions API is available
+  if not check_interactions_api_available():
+    print("\nERROR: Interactions API is not available in the current SDK.")
+    print("The interactions API requires a SDK version with this feature.")
+    print("To use the interactions API, ensure you have the SDK with")
+    print("interactions support installed (e.g., from private-python-genai).")
+    return False
+
+  test_agent = root_agent
+
+  runner = InMemoryRunner(
+      agent=test_agent,
+      app_name=APP_NAME,
+  )
+
+  # Create a new session
+  session = await runner.session_service.create_session(
+      user_id=USER_ID,
+      app_name=APP_NAME,
+  )
+  print(f"\nSession created: {session.id}")
+
+  try:
+    # Run all tests
+    await test_basic_text_generation(runner, session.id)
+    await test_function_calling(runner, session.id)
+    await test_multi_turn_conversation(runner, session.id)
+    await test_google_search_tool(runner, session.id)
+    await test_custom_function_tool(runner, session.id)
+
+    print("\n" + "=" * 60)
+    print("ALL TESTS PASSED (Interactions API)")
+    print("=" * 60)
+    return True
+
+  except AssertionError as e:
+    print(f"\nTEST FAILED: {e}")
+    return False
+  except Exception as e:
+    print(f"\nERROR: {e}")
+    import traceback
+
+    traceback.print_exc()
+    return False
+
+
+async def interactive_mode():
+  """Run in interactive mode for manual testing."""
+  # Check if interactions API is available
+  if not check_interactions_api_available():
+    print("\nERROR: Interactions API is not available in the current SDK.")
+    print("To use the interactions API, ensure you have the SDK with")
+    print("interactions support installed (e.g., from private-python-genai).")
+    return
+
+  print("\nInteractive mode with Interactions API")
+  print("Type 'quit' to exit, 'new' for a new session\n")
+
+  test_agent = agent.root_agent
+
+  runner = InMemoryRunner(
+      agent=test_agent,
+      app_name=APP_NAME,
+  )
+
+  session = await runner.session_service.create_session(
+      user_id=USER_ID,
+      app_name=APP_NAME,
+  )
+  print(f"Session created: {session.id}\n")
+
+  while True:
+    try:
+      user_input = input("You: ").strip()
+      if not user_input:
+        continue
+      if user_input.lower() == "quit":
+        break
+      if user_input.lower() == "new":
+        session = await runner.session_service.create_session(
+            user_id=USER_ID,
+            app_name=APP_NAME,
+        )
+        print(f"New session created: {session.id}\n")
+        continue
+
+      await call_agent_async(runner, USER_ID, session.id, user_input)
+
+    except KeyboardInterrupt:
+      break
+
+  print("\nGoodbye!")
+
+
+def main():
+  parser = argparse.ArgumentParser(
+      description="Test the Interactions API integration"
+  )
+  parser.add_argument(
+      "--mode",
+      choices=["test", "interactive"],
+      default="test",
+      help=(
+          "Run mode: 'test' runs automated tests, 'interactive' for manual"
+          " testing"
+      ),
+  )
+  parser.add_argument(
+      "--debug",
+      action="store_true",
+      help="Enable debug logging",
+  )
+
+  args = parser.parse_args()
+
+  if args.debug:
+    logs.setup_adk_logger(level=logging.DEBUG)
+  else:
+    logs.setup_adk_logger(level=logging.INFO)
+
+  start_time = time.time()
+
+  if args.mode == "test":
+    success = asyncio.run(run_all_tests())
+    if not success:
+      exit(1)
+
+  elif args.mode == "interactive":
+    asyncio.run(interactive_mode())
+
+  end_time = time.time()
+  print(f"\nTotal execution time: {end_time - start_time:.2f} seconds")
+
+
+if __name__ == "__main__":
+  main()

contributing/samples/manual_ollama_test/__init__.py

@@ -0,0 +1,15 @@
+# Copyright 2025 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from . import agent

contributing/samples/manual_ollama_test/agent.py

@@ -0,0 +1,39 @@
+# Copyright 2025 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from __future__ import annotations
+
+from google.adk.agents.llm_agent import LlmAgent
+from google.adk.agents.sequential_agent import SequentialAgent
+from google.adk.models.lite_llm import LiteLlm
+
+ollama_model = LiteLlm(model="ollama_chat/qwen2.5:7b")
+
+hello_agent = LlmAgent(
+    name="hello_step",
+    instruction="Say hello to the user. Be concise.",
+    model=ollama_model,
+)
+
+summarize_agent = LlmAgent(
+    name="summarize_step",
+    instruction="Summarize the previous assistant message in 5 words.",
+    model=ollama_model,
+)
+
+root_agent = SequentialAgent(
+    name="ollama_seq_test",
+    description="Two-step sanity check for Ollama LiteLLM chat.",
+    sub_agents=[hello_agent, summarize_agent],
+)

contributing/samples/migrate_session_db/main.py

@@ -32,7 +32,7 @@ logs.log_to_tmp_folder()
 async def main():
   app_name = 'migrate_session_db_app'
   user_id_1 = 'user1'
-  session_service = DatabaseSessionService('sqlite:///./sessions.db')
+  session_service = DatabaseSessionService('sqlite+aiosqlite:///./sessions.db')
   artifact_service = InMemoryArtifactService()
   runner = Runner(
       app_name=app_name,

contributing/samples/spanner/agent.py

@@ -18,6 +18,7 @@ from google.adk.agents.llm_agent import LlmAgent
 from google.adk.auth.auth_credential import AuthCredentialTypes
 from google.adk.tools.google_tool import GoogleTool
 from google.adk.tools.spanner.settings import Capabilities
+from google.adk.tools.spanner.settings import QueryResultMode
 from google.adk.tools.spanner.settings import SpannerToolSettings
 from google.adk.tools.spanner.spanner_credentials import SpannerCredentialsConfig
 from google.adk.tools.spanner.spanner_toolset import SpannerToolset
@@ -34,7 +35,10 @@ CREDENTIALS_TYPE = None
 
 
 # Define Spanner tool config with read capability set to allowed.
-tool_settings = SpannerToolSettings(capabilities=[Capabilities.DATA_READ])
+tool_settings = SpannerToolSettings(
+    capabilities=[Capabilities.DATA_READ],
+    query_result_mode=QueryResultMode.DICT_LIST,
+)
 
 if CREDENTIALS_TYPE == AuthCredentialTypes.OAUTH2:
   # Initialize the tools to do interactive OAuth

pyproject.toml

@@ -30,7 +30,7 @@ dependencies = [
   "anyio>=4.9.0, <5.0.0",                                    # For MCP Session Manager
   "authlib>=1.5.1, <2.0.0",                                  # For RestAPI Tool
   "click>=8.1.8, <9.0.0",                                    # For CLI tools
-  "fastapi>=0.115.0, <0.119.0",                              # FastAPI framework
+  "fastapi>=0.115.0, <0.124.0",                              # FastAPI framework
   "google-api-python-client>=2.157.0, <3.0.0",               # Google API client discovery
   "google-cloud-aiplatform[agent_engines]>=1.125.0, <2.0.0", # For VertexAI integrations, e.g. example store.
   "google-cloud-bigquery-storage>=2.0.0",
@@ -41,7 +41,7 @@ dependencies = [
   "google-cloud-spanner>=3.56.0, <4.0.0",                    # For Spanner database
   "google-cloud-speech>=2.30.0, <3.0.0",                     # For Audio Transcription
   "google-cloud-storage>=2.18.0, <4.0.0",                    # For GCS Artifact service
-  "google-genai>=1.51.0, <2.0.0",                            # Google GenAI SDK
+  "google-genai>=1.55.0, <2.0.0",                            # Google GenAI SDK
   "graphviz>=0.20.2, <1.0.0",                                # Graphviz for graph rendering
   "jsonschema>=4.23.0, <5.0.0",                              # Agent Builder config validation
   "mcp>=1.10.0, <2.0.0",                                     # For MCP Toolset
@@ -59,7 +59,7 @@ dependencies = [
   "requests>=2.32.4, <3.0.0",
   "sqlalchemy-spanner>=1.14.0",                              # Spanner database session service
   "sqlalchemy>=2.0, <3.0.0",                                 # SQL database ORM
-  "starlette>=0.46.2, <1.0.0",                               # For FastAPI CLI
+  "starlette>=0.49.1, <1.0.0",                               # For FastAPI CLI
   "tenacity>=9.0.0, <10.0.0",                                # For Retry management
   "typing-extensions>=4.5, <5",
   "tzlocal>=5.3, <6.0",                                      # Time zone utilities
@@ -116,7 +116,7 @@ test = [
   # go/keep-sorted start
   "a2a-sdk>=0.3.0,<0.4.0",
   "anthropic>=0.43.0",                            # For anthropic model tests
-  "crewai[tools];python_version>='3.10' and python_version<'3.12'",         # For CrewaiTool tests; chromadb/pypika fail on 3.12+
+  "crewai[tools];python_version>='3.11' and python_version<'3.12'",         # For CrewaiTool tests; chromadb/pypika fail on 3.12+
   "kubernetes>=29.0.0",                           # For GkeCodeExecutor
   "langchain-community>=0.3.17",
   "langgraph>=0.2.60, <0.4.8",                    # For LangGraphAgent
@@ -146,7 +146,7 @@ docs = [
 extensions = [
   "anthropic>=0.43.0",                          # For anthropic model support
   "beautifulsoup4>=3.2.2",                      # For load_web_page tool.
-  "crewai[tools];python_version>='3.10' and python_version<'3.12'",       # For CrewaiTool; chromadb/pypika fail on 3.12+
+  "crewai[tools];python_version>='3.11' and python_version<'3.12'",       # For CrewaiTool; chromadb/pypika fail on 3.12+
   "docker>=7.0.0",                              # For ContainerCodeExecutor
   "kubernetes>=29.0.0",                         # For GkeCodeExecutor
   "langgraph>=0.2.60, <0.4.8",                  # For LangGraphAgent

src/google/adk/a2a/converters/event_converter.py

@@ -140,6 +140,7 @@ def _get_context_metadata(
         ("custom_metadata", event.custom_metadata),
         ("usage_metadata", event.usage_metadata),
         ("error_code", event.error_code),
+        ("actions", event.actions),
     ]
 
     for field_name, field_value in optional_fields: