Merge remote-tracking branch 'upstream/main' into feat/add-metadata-parameter

Author: donggyun112

contributing/samples/adk_team/adk_pr_triaging_agent/README.md

@@ -1,6 +1,6 @@
 # ADK Pull Request Triaging Assistant
 
-The ADK Pull Request (PR) Triaging Assistant is a Python-based agent designed to help manage and triage GitHub pull requests for the `google/adk-python` repository. It uses a large language model to analyze new and unlabelled pull requests, recommend appropriate labels, assign a reviewer, and check contribution guides based on a predefined set of rules.
+The ADK Pull Request (PR) Triaging Assistant is a Python-based agent designed to help manage and triage GitHub pull requests for the `google/adk-python` repository. It uses a large language model to analyze new and unlabelled pull requests, recommend appropriate labels, and assign the component owner based on a predefined set of rules.
 
 This agent can be operated in two distinct modes:
 
@@ -16,7 +16,7 @@ This mode allows you to run the agent locally to review its recommendations in r
 ### Features
 
 - **Web Interface**: The agent's interactive mode can be rendered in a web browser using the ADK's `adk web` command.
-- **User Approval**: In interactive mode, the agent is instructed to ask for your confirmation before applying a label or posting a comment to a GitHub pull request.
+- **User Approval**: In interactive mode, the agent is instructed to ask for your confirmation before applying a label or assigning an owner to a GitHub pull request.
 
 ### Running in Interactive Mode
 
@@ -42,7 +42,7 @@ The GitHub workflow is configured to run on specific triggers:
 
 ### Automated Labeling
 
-When running as part of the GitHub workflow, the agent operates non-interactively. It identifies and applies the best label or posts a comment directly without requiring user approval. This behavior is configured by setting the `INTERACTIVE` environment variable to `0` in the workflow file.
+When running as part of the GitHub workflow, the agent operates non-interactively. It identifies and applies the best label and assigns the component owner directly without requiring user approval. This behavior is configured by setting the `INTERACTIVE` environment variable to `0` in the workflow file.
 
 ### Workflow Configuration
 

contributing/samples/adk_team/adk_pr_triaging_agent/agent.py

@@ -12,7 +12,6 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from pathlib import Path
 from typing import Any
 
 from adk_pr_triaging_agent.settings import GITHUB_BASE_URL
@@ -24,7 +23,6 @@
 from adk_pr_triaging_agent.utils import get_request
 from adk_pr_triaging_agent.utils import is_assignable
 from adk_pr_triaging_agent.utils import post_request
-from adk_pr_triaging_agent.utils import read_file
 from adk_pr_triaging_agent.utils import run_graphql_query
 from google.adk import Agent
 import requests
@@ -59,17 +57,13 @@
     "web": "wyf7107",
 }
 
-CONTRIBUTING_MD = read_file(
-    Path(__file__).resolve().parents[4] / "CONTRIBUTING.md"
-)
-
 APPROVAL_INSTRUCTION = (
-    "Do not ask for user approval for labeling, commenting, or assigning!"
+    "Do not ask for user approval for labeling or assigning!"
     " If you can't find appropriate labels for the PR, do not label it."
 )
 if IS_INTERACTIVE:
   APPROVAL_INSTRUCTION = (
-      "Only label, comment, or assign when the user approves the action!"
+      "Only label or assign when the user approves the action!"
   )
 
 
@@ -259,32 +253,6 @@ def assign_owner_to_pr(pr_number: int, label: str) -> dict[str, Any]:
   }
 
 
-def add_comment_to_pr(pr_number: int, comment: str) -> dict[str, Any]:
-  """Add the specified comment to the given PR number.
-
-  Args:
-    pr_number: the number of the GitHub pull request
-    comment: the comment to add
-
-  Returns:
-    The status of this request, with the applied comment when successful.
-  """
-  print(f"Attempting to add comment '{comment}' to issue #{pr_number}")
-
-  # Pull Request is a special issue in GitHub, so we can use issue url for PR.
-  url = f"{GITHUB_BASE_URL}/repos/{OWNER}/{REPO}/issues/{pr_number}/comments"
-  payload = {"body": comment}
-
-  try:
-    post_request(url, payload)
-  except requests.exceptions.RequestException as e:
-    return error_response(f"Error: {e}")
-  return {
-      "status": "success",
-      "added_comment": comment,
-  }
-
-
 def list_untriaged_pull_requests(pr_count: int) -> dict[str, Any]:
   """List open pull requests that need triaging.
 
@@ -348,8 +316,6 @@ def list_untriaged_pull_requests(pr_count: int) -> dict[str, Any]:
       - Get the pull request details.
       - Add a label to the pull request.
       - Assign the component owner (the shepherd) to the pull request.
-      - Check if the pull request is following the contribution guidelines.
-      - Add a comment to the pull request if it's not following the guidelines.
 
       **IMPORTANT: {APPROVAL_INSTRUCTION}**
 
@@ -367,59 +333,30 @@ def list_untriaged_pull_requests(pr_count: int) -> dict[str, Any]:
       - If it's about Model Context Protocol (e.g. MCP tool, MCP toolset, MCP session management etc.), label it with "mcp".
       - If you can't find an appropriate labels for the PR, follow the previous instruction that starts with "IMPORTANT:".
 
-      Here is the contribution guidelines:
-      `{CONTRIBUTING_MD}`
-
-      Here are the guidelines for checking if the PR is following the guidelines:
-      - The "statusCheckRollup" in the pull request details may help you to identify if the PR is following some of the guidelines (e.g. CLA compliance).
-
-      Here are the guidelines for the comment:
-      - **Be Polite and Helpful:** Start with a friendly tone.
-      - **Be Specific:** Clearly list only the sections from the contribution guidelines that are still missing.
-      - **Address the Author:** Mention the PR author by their username (e.g., `@username`).
-      - **Provide Context:** Explain *why* the information or action is needed.
-      - **Do not be repetitive:** If you have already commented on an PR asking for information, do not comment again unless new information has been added and it's still incomplete.
-      - **Identify yourself:** Include a bolded note (e.g. "Response from ADK Triaging Agent") in your comment to indicate this comment was added by an ADK Answering Agent.
-
-      **Example Comment for a PR:**
-      > **Response from ADK Triaging Agent**
-      >
-      > Hello @[pr-author-username], thank you for creating this PR!
-      >
-      > This PR is a bug fix, could you please associate the github issue with this PR? If there is no existing issue, could you please create one?
-      >
-      > In addition, could you please provide logs or screenshot after the fix is applied?
-      >
-      > This information will help reviewers to review your PR more efficiently. Thanks!
-
       # 4. Steps
       - If you are asked to find pull requests that need triaging, use `list_untriaged_pull_requests` first.
       - For each pull request to be triaged:
         - Call the `get_pull_request_details` tool to get the details of the PR.
-        - Skip the PR (i.e. do not label or comment) if any of the following is true:
+        - Skip the PR (i.e. do not label) if any of the following is true:
           - the PR is closed
           - the PR is labeled with "google-contributor"
           - the PR is already labelled with the above labels (e.g. "documentation", "services", "tools", etc.).
-        - Check if the PR is following the contribution guidelines.
-          - If it's not following the guidelines, recommend or add a comment to the PR that points to the contribution guidelines (https://github.com/google/adk-python/blob/main/CONTRIBUTING.md).
-          - If it's following the guidelines, recommend or add a label to the PR.
+        - Recommend or add a label to the PR.
         - After you add a component label, assign the component owner (the shepherd) to the PR:
           - Call `assign_owner_to_pr` with the same label you applied.
           - Skip assignment if the PR already has an assignee.
-          - If the tool reports the owner is not assignable, just note it; do not comment about it.
+          - If the tool reports the owner is not assignable, just note it.
 
       # 5. Output
       Present the following in an easy to read format highlighting PR number and your label.
       - The PR summary in a few sentence
       - The label you recommended or added with the justification
       - The owner you assigned (or why you did not)
-      - The comment you recommended or added to the PR with the justification
     """,
     tools=[
         list_untriaged_pull_requests,
         get_pull_request_details,
         add_label_to_pr,
         assign_owner_to_pr,
-        add_comment_to_pr,
     ],
 )

contributing/samples/live/live_bidi_streaming_tools_agent/agent.py

@@ -22,7 +22,15 @@
 
 
 async def monitor_stock_price(stock_symbol: str) -> AsyncGenerator[str, None]:
-  """This function will monitor the price for the given stock_symbol in a continuous, streaming and asynchronously way."""
+  """Starts a background monitor for the price of the given stock_symbol.
+
+  Call this function ONLY ONCE to initiate monitoring. Once started, it runs
+  continuously in the background and automatically streams price alerts.
+
+  CRITICAL: Do NOT call this function again to "check" or "poll" for updates.
+  Simply wait for the background task to yield new values and report them.
+  Calling this again while running will launch a duplicate background task.
+  """
   print(f"Start monitor stock price for {stock_symbol}!")
 
   # Let's mock stock price change.
@@ -51,13 +59,21 @@ async def monitor_stock_price(stock_symbol: str) -> AsyncGenerator[str, None]:
 async def monitor_video_stream(
     input_stream: LiveRequestQueue,
 ) -> AsyncGenerator[str, None]:
-  """Monitor how many people are in the video streams."""
+  """Starts a background monitor for the video stream.
+
+  Call this function ONLY ONCE to initiate monitoring. Once started, it runs
+  continuously in the background and automatically streams updates back to
+  you whenever the person count changes.
+
+  CRITICAL: Do NOT call this function again to "check" or "poll" for updates.
+  Simply wait for the background task to yield new values and report them.
+  Calling this again while running will launch a duplicate background task.
+  """
   from google.genai import Client
 
   print("start monitor_video_stream!")
-  from google.genai import Client
 
-  client = Client(enterprise=False)
+  client = Client()
   prompt_text = (
       "Count the number of people in this image. Just respond with a numeric"
       " number."
@@ -90,6 +106,7 @@ async def monitor_video_stream(
 
       # Call the model to generate content based on the provided image and prompt
       response = client.models.generate_content(
+          model="gemini-2.5-flash",
           contents=contents,
           config=genai_types.GenerateContentConfig(
               system_instruction=(
@@ -99,12 +116,12 @@ async def monitor_video_stream(
               )
           ),
       )
+      new_count = response.candidates[0].content.parts[0].text.strip()
       if not last_count:
-        last_count = response.candidates[0].content.parts[0].text
-      elif last_count != response.candidates[0].content.parts[0].text:
-        last_count = response.candidates[0].content.parts[0].text
-        yield response
-        print("response:", response)
+        last_count = new_count
+      elif last_count != new_count:
+        last_count = new_count
+        yield new_count
 
     # Wait before checking for new images
     await asyncio.sleep(0.5)
@@ -135,6 +152,11 @@ def stop_streaming(function_name: str):
       You can use monitor_video_stream function to do that. When monitor_video_stream
       returns the alert, you should tell the users.
       When users want to monitor a stock price, you can use monitor_stock_price.
+      CRITICAL: Only call the monitor tools (monitor_video_stream, monitor_stock_price) at most once per request.
+      Once called, these tools run continuously in the background. Do NOT call them again to "poll" or "check" for updates.
+      Instead, simply wait for the background tool to stream a new message/alert to you, and then report that alert to the user.
+      Calling the tool again while it is already running will cause duplicate tasks and errors.
+      If you need to stop a monitor, call stop_streaming. Only after stopping can you call the monitor tool again if needed.
       Don't ask too many questions. Don't be too talkative.
     """,
     tools=[

src/google/adk/cli/api_server.py

@@ -742,7 +742,10 @@ async def get_runner_async(self, app_name: str) -> Runner:
       return self.runner_dict[app_name]
 
     # Create new runner
-    agent_or_app = self.agent_loader.load_agent(app_name)
+    try:
+      agent_or_app = self.agent_loader.load_agent(app_name)
+    except ValueError as ve:
+      raise HTTPException(status_code=404, detail=str(ve)) from ve
 
     if self.default_llm_model:
       from .cli import _override_default_llm_model
@@ -1162,7 +1165,10 @@ async def get_adk_app_info(app_name: str) -> AppInfo:
                 " mode."
             ),
         )
-      agent_or_app = self.agent_loader.load_agent(app_name)
+      try:
+        agent_or_app = self.agent_loader.load_agent(app_name)
+      except ValueError as ve:
+        raise HTTPException(status_code=404, detail=str(ve)) from ve
       root_agent = self._get_root_agent(agent_or_app)
       if isinstance(root_agent, LlmAgent):
         return AppInfo(

src/google/adk/flows/llm_flows/contents.py

@@ -387,7 +387,7 @@ def _build_task_input_user_content(
         try:
           import json as _json
 
-          text = _json.dumps(dict(fc.args))
+          text = _json.dumps(dict(fc.args), ensure_ascii=False)
         except (TypeError, ValueError):
           text = str(fc.args)
         parts = [types.Part(text=text)]