20 Pull requests merged by 4 people

• feat: Add custom_metadata to VertexAiSessionService event handling

  #1794 merged Jul 4, 2025

• fix: Replace Event ID generation with UUID4 to prevent collisions

  #1778 merged Jul 4, 2025

• fix: Remove duplicate options from adk deploy

  #1767 merged Jul 3, 2025

• chore: update agents.md to include versioning strategy

  #1765 merged Jul 2, 2025

• fix: issue 1641 database session service leaks events

  #1644 merged Jul 2, 2025

• feat: Add Safety evaluator metric

  #1737 merged Jul 2, 2025

• chore: Update ResponseEvaluator to use newer version of Eval SDK

  #1724 merged Jul 2, 2025

• doc: Update README.md

  #1744 merged Jul 2, 2025

• chore: Add tenacity into project.toml

  #1755 merged Jul 2, 2025

• Fix: Handle unexpected 'parameters' argument in FunctionTool.run_async

  #1656 merged Jul 2, 2025

• chore: Update the retry logic of create session polling

  #1743 merged Jul 2, 2025

• fix: continue the action when grep -L returns exit status 1

  #1677 merged Jul 1, 2025

• chore: Add util to build our llms.txt and llms-full.txt files

  #1709 merged Jul 1, 2025

• feat: BaseEvalService declaration and surrounding data models

  #1353 merged Jul 1, 2025

• fix: Make sure each partial event has different timestamp

  #1729 merged Jul 1, 2025

• chore: update adk web

  #1715 merged Jun 30, 2025

• fix: Avoid pydantic.ValidationError when the model stream returns empty final chunk

  #1706 merged Jun 30, 2025

• chore: Create an example for multi agent live streaming

  #1716 merged Jun 30, 2025

• docs: Add AGENTS.md to provide relevant project context for the Gemini CLI

  #1708 merged Jun 28, 2025

• chore: refactor the ADK Triaging Agent to make the code easier to read

  #1508 merged Jun 27, 2025

15 Pull requests opened by 8 people

• feat: add agent.clone() method

  #1707 opened Jun 27, 2025

• chore: create an initial prototype agent to answer Github issue questions

  #1741 opened Jul 1, 2025

• ADK changes

  #1745 opened Jul 2, 2025

• feat: add agent.clone() method

  #1747 opened Jul 2, 2025

• fix(llm): handle ConnectionClosedOK exception in _receive_from_model test

  #1752 opened Jul 2, 2025

• feat: Add implementation of BaseEvalService that runs evals locally

  #1756 opened Jul 2, 2025

• fix: Remove end_invocation check to ensure after_agent callbacks execute

  #1761 opened Jul 2, 2025

• fix(code-executor): resolve deepcopy recursion in VertexAiCodeExecutor

  #1762 opened Jul 2, 2025

• Fix: Add missing Union import in runners.py

  #1763 opened Jul 2, 2025

• feat: Create PR agent for ADK github repo

  #1764 opened Jul 2, 2025

• fix: Handle HttpResponse objects in VertexAI session service API calls (fix #1772)

  #1773 opened Jul 3, 2025

• feat: Feature/support tool callbacks in live mode

  #1774 opened Jul 3, 2025

• test: Adds test for streaming + function calls

  #1780 opened Jul 3, 2025

• test: Add tests for live streaming configs

  #1781 opened Jul 3, 2025

• Add support for including history windowin current turn contents

  #1790 opened Jul 4, 2025

30 Issues closed by 8 people

• When calling adk via Fast API, if agent is asked to do intermediate responses makes ADK crash

  #719 closed Jul 4, 2025

• Trouble deploying agent to cloud run.

  #1739 closed Jul 3, 2025

• Setting Env variables deploying to Agent Engine

  #693 closed Jul 2, 2025

• When querying events during the get_session process, there is a logical error.

  #1341 closed Jul 2, 2025

• Unable to deploy basic agent engine using MCPToolset to Vertex Agent Engine

  #1024 closed Jul 2, 2025

• using VertexAiRagMemoryService in AdkApp

  #432 closed Jul 2, 2025

• DatabaseSessionService get_session does not filter events with user_id

  #1641 closed Jul 2, 2025

• Streaming implementation question: using get_fast_api vs using streaming implementation page

  #1225 closed Jul 2, 2025

• Tool or sub-agent returns in a streaming manner.

  #1345 closed Jul 2, 2025

• Cannot add subagents and AgentTools in streaming mode. TypeError asyncio extra_headers

  #1348 closed Jul 2, 2025

• ❓ Question: Integrating ADK with CopilotKit via AG-UI protocol

  #1229 closed Jul 2, 2025

• The operation process of the intelligent agent

  #1399 closed Jul 2, 2025

• Change timeout for MCP Server (ADK enforces 5 seconds)

  #1463 closed Jul 2, 2025

• New Gemini LIVE model naming got blocked by Google Search tool use

  #1512 closed Jul 2, 2025

• Please write api breaking changes about adk api_server `create_session` api in the release note (between `1.3.0` and `1.4.1`)

  #1564 closed Jul 2, 2025

• function call param resolve incorrectly

  #287 closed Jul 2, 2025

• Facing issue while executing tool calls

  #283 closed Jul 2, 2025

• mcp tool call issue: not allowing dictionary input

  #281 closed Jul 2, 2025

• Get RecursionError when running using custom agents

  #461 closed Jul 2, 2025

• Multiple oauth2 authenticated tools calls in single request results in "invalid grant"

  #390 closed Jul 2, 2025

• Agent exits early

  #409 closed Jul 2, 2025

• `from google.adk.agents import Agent` redirects to class LlmAgent

  #1158 closed Jul 2, 2025

• Using the code_executor parameter in Agent

  #1620 closed Jul 2, 2025

• Parallel Workflow with run_live()

  #1736 closed Jul 1, 2025

• Can a RemoteAgent class be provided to encapsulate a remote A2A server?

  #1710 closed Jun 30, 2025

• Hello! How to use adk web with non google api model? (Deepinfra or local)

  #1702 closed Jun 30, 2025

• DatabaseSessionService() doesn't store data to db

  #1587 closed Jun 30, 2025

• VertexAISession not handling HttpResponses properly

  #1539 closed Jun 28, 2025

• litellm PerplexityException

  #1701 closed Jun 28, 2025

• ‼️ json.decoder.JSONDecodeError: Extra data: line 3 column 1 (char 639)

  #1705 closed Jun 28, 2025

40 Issues opened by 34 people

• Global Session Storage

  #1793 opened Jul 4, 2025

• [QUESTION] AgentEvaluator.evaluate_eval_set

  #1792 opened Jul 4, 2025

• [bug] Potential MCP auth slowness

  #1791 opened Jul 4, 2025

• QUESTION: It seems that LiteLLM does not handle the reasoning_content of the model.

  #1788 opened Jul 4, 2025

• if sub_agent name use chinese,ADK use it with a tool.and get a error.

  #1787 opened Jul 4, 2025

• ERROR: transfer_to_agent() got an unexpected keyword argument 'prompt', sometime argument 'user_request', argument

  #1786 opened Jul 4, 2025

• Windows log symlink creation fails without admin rights; suggest configurable log dir and graceful fallback

  #1785 opened Jul 4, 2025

• Callbacks are not triggered when using Agent as a Tool

  #1783 opened Jul 4, 2025

• QUESTION: adk web support multi-agents with different models?

  #1782 opened Jul 3, 2025

• MCP Resources in addition to Tools

  #1779 opened Jul 3, 2025

• `VertexAiSessionService#get_session` raise an error when a session has many events (`list_events_api_response` has `nextPageToken`)

  #1772 opened Jul 3, 2025

• SequentialAgent replying mid-chain

  #1770 opened Jul 3, 2025

• [Feature Request] Support tool callback for live(bidi)

  #1768 opened Jul 3, 2025

• Compatibility Issue with ForwardingArtifactService and Custom Artifact Services

  #1766 opened Jul 2, 2025

• Agent Engine Session URI Construction Incorrectly Relies on `GOOGLE_CLOUD_LOCATION`

  #1760 opened Jul 2, 2025

• Master issue - live : support callbacks for live(bidi)

  #1759 opened Jul 2, 2025

• Master Issue - Live: support session and events

  #1757 opened Jul 2, 2025

• Recommended way to handle files not directly supported by Gemini

  #1751 opened Jul 2, 2025

• synchronous psycopg2 dependency in postgres database sessionservice , while adk is async all the way

  #1750 opened Jul 2, 2025

• After using ADK in depth, it is recommended to support the thinking model and bug feedback

  #1749 opened Jul 2, 2025

• Incorrect Role Assignment in Multi-Agent Mode Causes Context Misinterpretation

  #1748 opened Jul 2, 2025

• Custom Invocation Context propogation when using AgentTool

  #1746 opened Jul 2, 2025

• Docs: Typo in Configuring with "Configuring Tools with Authentication" sample code

  #1740 opened Jul 1, 2025

• LLMAgent loses context/state after tool execution in Custom Agent

  #1738 opened Jul 1, 2025

• VertexAIRagRetrieval error when using in SequentialAgent in streaming mode.

  #1734 opened Jul 1, 2025

• The last_update_time provided in the session object '2025-07-01 18:59:42' is earlier than the update_time in the storage_session '2025-07-01 18:59:56'. Please check if it is a stale session

  #1733 opened Jul 1, 2025

• Add more visibility into how Agent performed when called as a tool

  #1732 opened Jul 1, 2025

• First describe the action to be performed, then invoke the tool, like cursor

  #1731 opened Jul 1, 2025

• Can't deploy adk with VertexAiCodeExecutor

  #1730 opened Jul 1, 2025

• New/updated features are not visible to old customer

  #1728 opened Jul 1, 2025

• Cannot deploy ADK agent with MCPToolset to Agent Engine - "TypeError: cannot pickle 'TextIOWrapper' instances"

  #1727 opened Jul 1, 2025

• README Link giving 404 - page not found

  #1726 opened Jul 1, 2025

• StreamingTool fails to generate proper schema

  #1725 opened Jul 1, 2025

• Design Proposal: PatchEvent for Session Pruning (Relates to #1013 & #1588)

  #1723 opened Jul 1, 2025

• Mismatch in image_url and video_url parameter format for vision models

  #1722 opened Jun 30, 2025

• LangchainTools in `run_live` throw Union typing error error

  #1720 opened Jun 30, 2025

• UnsupportedParamsError when using litellm with perplexity/sonar model

  #1718 opened Jun 30, 2025

• Audio distortion/delay upon sub-agent transfer in live models.

  #1717 opened Jun 30, 2025

• McpError: Connection closed for youtube_transcript_api

  #1713 opened Jun 30, 2025

• Streaming-Tools: saving artifacts does not produce artifact_delta event

  #1711 opened Jun 29, 2025

201 Unresolved conversations

Sometimes conversations happen on old items that aren’t yet closed. Here is a list of all the Issues and Pull Requests with unresolved conversations.

• Feature Request: Adopt llms.txt and llms-full.txt for AI Agent Consumption

  #1694 commented on Jun 27, 2025 • 0 new comments

• Reading text from scanned PDF images

  #1337 commented on Jun 28, 2025 • 0 new comments

• Feature Request: Return session state in List Sessions API in `adk api_server`

  #781 commented on Jun 29, 2025 • 0 new comments

• support for context caching

  #211 commented on Jun 30, 2025 • 0 new comments

• ApplicationIntegrationToolset.get_tools' was never awaited

  #1037 commented on Jun 30, 2025 • 0 new comments

• Provide an option to remove events from the session

  #1013 commented on Jul 1, 2025 • 0 new comments

• Support for cache_control in litellm

  #994 commented on Jul 1, 2025 • 0 new comments

• Customization in LLM request formation

  #1704 commented on Jul 2, 2025 • 0 new comments

• Error in event_generator: Unterminated string starting at

  #1666 commented on Jul 2, 2025 • 0 new comments

• "adk deploy agent_engine" command fails if there is a package installed with the name "agent"

  #1652 commented on Jul 2, 2025 • 0 new comments

• How to make the agent not output "For Context: xxx"

  #1639 commented on Jul 2, 2025 • 0 new comments

• TypeError: transfer_to_agent() got an unexpected keyword argument 'parameters'

  #1637 commented on Jul 2, 2025 • 0 new comments

• int | None union syntax fails to parse in function parameters despite documented support

  #1634 commented on Jul 2, 2025 • 0 new comments

• Option to iterate over an array and generated array response

  #1610 commented on Jul 2, 2025 • 0 new comments

• Google ADK 1.2.1 incompatible with Python 3.13- causes runtime failures

  #1593 commented on Jul 2, 2025 • 0 new comments

• Context buffer grows unboundedly—every past message is retained

  #1588 commented on Jul 2, 2025 • 0 new comments

• Delayed function responses (e.g. from auth) can result in INVALID_ARGUMENT due to unequal number of response parts to call parts

  #1583 commented on Jul 2, 2025 • 0 new comments

• Error INVALID_ARGUMENT: Please ensure that function call turn comes immediately after a user turn or after a function response turn

  #1581 commented on Jul 2, 2025 • 0 new comments

• Stop execution mid-way based on output_key on a sub-agent in SequentialAgent

  #1570 commented on Jul 2, 2025 • 0 new comments

• High latency observed in Agent (Root, subagents) particular actions attached to runner

  #1566 commented on Jul 2, 2025 • 0 new comments

• Feature: Secure Link Masking/Unmasking in Agent Responses

  #1565 commented on Jul 2, 2025 • 0 new comments

• AgentEngines update() fails with mandatory pickled_object_gcs_uri which does not seem to be required in ADK AgenEngines update()

  #1559 commented on Jul 2, 2025 • 0 new comments

• Agent Executing Tool endlessly

  #1555 commented on Jul 2, 2025 • 0 new comments

• 'LlmRequest' object has no attribute 'content'

  #1534 commented on Jul 2, 2025 • 0 new comments

• invocation_id mismatch in merge_parallel_function_response_events creates new event_id instead of reusing original invocation_id

  #1531 commented on Jul 2, 2025 • 0 new comments

• Pass the value of GOOGLE_CLOUD_LOCATION and GOOGLE_GENAI_USE_VERTEXAI while creating the object of Ruuner

  #1491 commented on Jul 2, 2025 • 0 new comments

• Error when stacking agents: Prompt leakage for the master agent

  #1460 commented on Jul 2, 2025 • 0 new comments

• MCPTimeout error inside linux container

  #1453 commented on Jul 2, 2025 • 0 new comments

• Enable development workflows and tooling to leverage virtual environments through `uv run`

  #1438 commented on Jul 2, 2025 • 0 new comments

• Allow users to define GOOGLE_CLOUD_PROJECT, GOOGLE_CLOUD_LOCATION, GOOGLE_GENAI_USE_VERTEXAI in code

  #1417 commented on Jul 2, 2025 • 0 new comments

• VideoAnalysisAgent returns incorrect video summary when provided with a YouTube link

  #1409 commented on Jul 2, 2025 • 0 new comments

• Deterministic Method to Exit a LoopAgent and Continue a Parent SequentialAgent

  #1376 commented on Jul 2, 2025 • 0 new comments

• Bug: The "thought: true" flag is lost in conversation history on subsequent requests

  #1338 commented on Jul 2, 2025 • 0 new comments

• Ability to customise API endpoints?

  #1326 commented on Jul 2, 2025 • 0 new comments

• Start from root agent after any sub-agent is skipped by before agent callback

  #1311 commented on Jul 2, 2025 • 0 new comments

• Conditional routing

  #1270 commented on Jul 2, 2025 • 0 new comments

• Add Agents to Tree Dynamically

  #1249 commented on Jul 2, 2025 • 0 new comments

• I am trying to access the data from State in a multi-agent system, however the last state update is missing

  #1238 commented on Jul 2, 2025 • 0 new comments

• pydantic_core._pydantic_core.ValidationError

  #1227 commented on Jul 2, 2025 • 0 new comments

• Dynamic Filter Configuration for Vector Search Grounding in LLM Agent

  #1215 commented on Jul 2, 2025 • 0 new comments

• Add a way to send files when root_agent is deployed to Google's cloud "Agent Engine"

  #1203 commented on Jul 2, 2025 • 0 new comments

• Feature Request: Per-Agent RunConfig Support for Individual Sub-Agents and Coordinator Agents

  #1167 commented on Jul 2, 2025 • 0 new comments

• Bug: state_delta from parallel AgentTool calls not aggregated

  #1153 commented on Jul 2, 2025 • 0 new comments

• Consistently running into 500 errors

  #1138 commented on Jul 2, 2025 • 0 new comments

• consitency: cli-flags

  #1118 commented on Jul 2, 2025 • 0 new comments

• Agent Raw response thought instead of final result

  #1109 commented on Jul 2, 2025 • 0 new comments

• `skip_summarization` does not work as intended when using `AgentTool` within nested agent setup

  #1103 commented on Jul 2, 2025 • 0 new comments

• Unsupported file uri: INVALID_ARGUMENT run_sse

  #1097 commented on Jul 2, 2025 • 0 new comments

• Set number of concurrency to `ParallelAgent`

  #1081 commented on Jul 2, 2025 • 0 new comments

• Service URL cannot serve traffic when deployed in Cloud Run using adk command & cannot use agent variable names other than "root_agent".

  #974 commented on Jul 2, 2025 • 0 new comments

• Allow configuration of client-side timeout to prevent 499 CANCELLED

  #475 commented on Jul 2, 2025 • 0 new comments

• feat: MergeAgent to handle output keys and handle reducer logic for LoopAgent and ParallelAgent

  #404 commented on Jul 2, 2025 • 0 new comments

• Context Bloat with Agent Passoff

  #1047 commented on Jul 2, 2025 • 0 new comments

• Add timestamp to contents generated by events

  #1074 commented on Jul 2, 2025 • 0 new comments

• Passing artifacts to AgentTool

  #992 commented on Jul 2, 2025 • 0 new comments

• OAuth for agents deployed to Agent Engine

  #1067 commented on Jul 2, 2025 • 0 new comments

• Tool call response is ignored — assistant does not produce final message after tool response

  #242 commented on Jul 2, 2025 • 0 new comments

• Generate API references for InvocationContext

  #291 commented on Jul 2, 2025 • 0 new comments

• Pydantic ValidationError (extra_forbidden for properties.resource.$ref) when using LangchainTool with google-adk

  #310 commented on Jul 2, 2025 • 0 new comments

• Crash while exiting loop agent in adk web

  #501 commented on Jul 2, 2025 • 0 new comments

• Issue with session state persistence using ToolContext.

  #562 commented on Jul 2, 2025 • 0 new comments

• before_agent_callback is not working as expected

  #1567 commented on Jul 2, 2025 • 0 new comments

• bug: `built_in_code_execution` output issue

  #180 commented on Jul 2, 2025 • 0 new comments

• Google Search Tool does not return grounding metadata in ADK Runner events

  #1693 commented on Jul 2, 2025 • 0 new comments

• after_agent_callback is not called when stop processing events after is_final_response is True

  #1695 commented on Jul 2, 2025 • 0 new comments

• Feature Request: Add an Endpoint to Explicitly Stop/Terminate a Conversation

  #1621 commented on Jul 2, 2025 • 0 new comments

• "Session not found" error

  #1575 commented on Jul 2, 2025 • 0 new comments

• Add fixes for InMemorySessionService behavior across long-lived agent requests

  #1569 commented on Jul 2, 2025 • 0 new comments

• DatabaseSessionService has race condition with `create_all()` when running multiple instances/replicas

  #1561 commented on Jul 2, 2025 • 0 new comments

• Session can't continue after tool error or stopping stream

  #1484 commented on Jul 2, 2025 • 0 new comments

• How is this possible (No tool call before tool response)?

  #1434 commented on Jul 2, 2025 • 0 new comments

• Add SQLAlchemy Engine Parameter to DatabaseSessionService for Custom Engine Support

  #1397 commented on Jul 2, 2025 • 0 new comments

• Use NotFoundError instead of ValueError when session not found

  #1383 commented on Jul 2, 2025 • 0 new comments

• Add new argument artifact_storage_service in get_fast_api_app function to pass Artifactory services instances

  #1367 commented on Jul 2, 2025 • 0 new comments

• AgentEngine fails to create session

  #1366 commented on Jul 2, 2025 • 0 new comments

• GCS artifact storage should be async

  #1346 commented on Jul 2, 2025 • 0 new comments

• Save Artifact via FastAPI API

  #1301 commented on Jul 2, 2025 • 0 new comments

• Add `session_db_kwargs` to FastAPI App

  #1287 commented on Jul 2, 2025 • 0 new comments

• greenlet_spawn has not been called; can't call await_only() here. Was IO attempted in an unexpected place? (Background on this error at: https://sqlalche.me/e/20/xd2s)

  #1276 commented on Jul 2, 2025 • 0 new comments

• Separate session for each agent in a multi-agent design

  #1254 commented on Jul 2, 2025 • 0 new comments

• Session often not accessible to subsequent agents, timing issue?

  #1119 commented on Jul 2, 2025 • 0 new comments

• Add endpoint to update session state

  #1117 commented on Jul 2, 2025 • 0 new comments

• Missing required tool parameters for Claude (AnthropicVertex)

  #1692 commented on Jul 2, 2025 • 0 new comments

• An internal error has occurred. Please retry or report in https://developers.generativeai.google/guide/troubleshooting'

  #1691 commented on Jul 2, 2025 • 0 new comments

• BUG: Function calls appear in the output as opposed to the output itself

  #1648 commented on Jul 2, 2025 • 0 new comments

• Tool Calling with Gemini Pro and Gemini Flash

  #1605 commented on Jul 2, 2025 • 0 new comments

• Tool calling Issue with Gemini models as ADK Agent model

  #1600 commented on Jul 2, 2025 • 0 new comments

• Claude MODELS Max token not fetching from generation config

  #1584 commented on Jul 2, 2025 • 0 new comments

• Function is not found in the tools_dict

  #1532 commented on Jul 2, 2025 • 0 new comments

• Removing a param from a tool does not update agent's internal understanding of the required param

  #1439 commented on Jul 2, 2025 • 0 new comments

• Weird final Model Response even though function output is correct. Any help is appreciated.

  #1335 commented on Jul 2, 2025 • 0 new comments

• how to add trim messages

  #1265 commented on Jul 2, 2025 • 0 new comments

• ERROR:opentelemetry.context:Failed to detach context

  #1670 commented on Jul 2, 2025 • 0 new comments

• Efficient Large Context Window Handling for Gemini Models

  #1246 commented on Jul 2, 2025 • 0 new comments

• Tool call fails in ADK

  #1168 commented on Jul 2, 2025 • 0 new comments

• Add inference providers directly

  #1151 commented on Jul 2, 2025 • 0 new comments

• Region needs to be configurable on every model

  #1120 commented on Jul 2, 2025 • 0 new comments

• `llm_request.model` not respected when mixing model providers due to agent-based LLM resolution

  #1116 commented on Jul 2, 2025 • 0 new comments

• Support of: gemini-2.5-flash-preview-tts [ 500 INTERNAL ]

  #1113 commented on Jul 2, 2025 • 0 new comments

• Add native Open-WebUI support as a backend for Google ADK

  #1096 commented on Jul 2, 2025 • 0 new comments

• Gemini Integration with ADK: Model Stability & None Response Issue

  #1088 commented on Jul 2, 2025 • 0 new comments

• Issue in accessing Litellm custom endpoint

  #985 commented on Jul 2, 2025 • 0 new comments

• how to set base url for Gemini model

  #489 commented on Jul 2, 2025 • 0 new comments

• ADK/LiteLLM: BadRequestError with Claude 3.7 on Vertex When Using Tools and thinking Feature Enabled

  #265 commented on Jul 2, 2025 • 0 new comments

• Feature Request: Create a Status Messenger Agent Tool for Real-time Observability

  #1697 commented on Jul 2, 2025 • 0 new comments

• Feature Request: Create an Agent Tool for GCP Pub/Sub Event Publishing

  #1696 commented on Jul 2, 2025 • 0 new comments

• Failed to publish integration version. Error: Event parameter `Task_1_dynamicAuthConfig` accessed from Task number 1 (GenericConnectorTaskImpl) is of the incorrect type or does not exist.

  #1628 commented on Jul 2, 2025 • 0 new comments

• Gcloud file not being found in Windows Installing while using adk deploy

  #1597 commented on Jul 2, 2025 • 0 new comments

• Persistent TypeError during FunctionTool initialization due to argument parsing inconsistencies

  #1592 commented on Jul 2, 2025 • 0 new comments

• RestApiTool：_get_declaration method

  #1578 commented on Jul 2, 2025 • 0 new comments

• Use Dependabot to keep GitHub Actions versions up to date

  #1548 commented on Jul 2, 2025 • 0 new comments

• Object of type AnyUrl is not JSON serializable while reading github file using MCP Tool

  #1540 commented on Jul 2, 2025 • 0 new comments

• Improve ToolFunction to recognize any variable typed ToolContext rather than only the hardcoded "tool_context"

  #1530 commented on Jul 2, 2025 • 0 new comments

• More Production Grade Strategy to Access User Identity in Tools

  #1522 commented on Jul 2, 2025 • 0 new comments

• Improve default argument tool use warning message

  #1482 commented on Jul 2, 2025 • 0 new comments

• ADK deployment VertexAiSearchTool not working; generates PERMISSION_DENIED error

  #1476 commented on Jul 2, 2025 • 0 new comments

• Client-credentials grant still opens browser & appends `response_type=code`

  #1475 commented on Jul 2, 2025 • 0 new comments

• adk deploy: handle non-python dependencies (npx, typescript mpcs, etc)

  #1473 commented on Jul 2, 2025 • 0 new comments

• fast_api.py:804 - Error in event_generator: name 'Optional' is not defined

  #1437 commented on Jul 2, 2025 • 0 new comments

• MCP Tools: No mechanism to pass user JWT token through ADK session.state to MCP server context

  #1402 commented on Jul 2, 2025 • 0 new comments

• Critical: Persistent TypeError in FunctionTool.__init__() across various argument patterns in google-adk==1.2.1, compounded by API documentation inconsistencies.

  #1331 commented on Jul 2, 2025 • 0 new comments

• Improve error handling in tools provided via BigQueryToolset

  #1324 commented on Jul 2, 2025 • 0 new comments

• Local MCP wouldn't start on windows / git bash

  #1321 commented on Jul 2, 2025 • 0 new comments

• Add mermaid diagram export capabilities for agent workflow understandability

  #1309 commented on Jul 2, 2025 • 0 new comments

• Add function to enable streamlined human user prompt input and agentic response stream

  #1284 commented on Jul 2, 2025 • 0 new comments

• MCPTOOLSET issue on connecting new mcp server with google ADK

  #1277 commented on Jul 2, 2025 • 0 new comments

• Support tool-calling for LLMs with coding skills

  #1273 commented on Jul 2, 2025 • 0 new comments

• API Calls using OPENAPI toolset fails with SSL verification failure

  #1272 commented on Jul 2, 2025 • 0 new comments

• Intermittent 400 Bad Request & anyio.BrokenResourceError with MCPToolset.get_tools() using StreamableHTTPServerParams, leading to asyncio.CancelledError on retry

  #1267 commented on Jul 2, 2025 • 0 new comments

• Add dataStoreSpecs[] and boost_spec to Vertex AI Search

  #1259 commented on Jul 2, 2025 • 0 new comments

• ADK MCP Client should support the get images method

  #1221 commented on Jul 2, 2025 • 0 new comments

• ADK MCP Client should support getting context from an MCP server

  #1220 commented on Jul 2, 2025 • 0 new comments

• GmailToolset OAuth redirect_uri malformed with appended flowName parameter

  #1209 commented on Jul 2, 2025 • 0 new comments

• Empty Google Search Grounding

  #1193 commented on Jul 2, 2025 • 0 new comments

• MALFORMED_FUNCTION_CALL when using OpenAPIToolSet

  #1192 commented on Jul 2, 2025 • 0 new comments

• after_tool_callback not triggering for VertexAiSearchTool in Google ADK

  #1121 commented on Jul 2, 2025 • 0 new comments

• When adk calls the MCP tool based on nodejs, it incorrectly identifies tool names with '-' as' _ ', resulting in errors

  #1115 commented on Jul 2, 2025 • 0 new comments

• Feature requrest for starting toolbox server with both prebuilt tools and tools file

  #1106 commented on Jul 2, 2025 • 0 new comments

• MCPToolset Timeout Error

  #1086 commented on Jul 2, 2025 • 0 new comments

• Investigating unexpected AttributeError with AgentTool and argument handling discrepancy in streaming multimodal agents.

  #1084 commented on Jul 2, 2025 • 0 new comments

• Pydantic ValidationError: Schema defs Input should be a valid dictionary with MCPToolset and Notion MCP Server (ADK 1.1.1, Pydantic 2.11.5)

  #1082 commented on Jul 2, 2025 • 0 new comments

• MCPToolset (BrowserMCP) Times Out When Used Inside LoopAgent

  #1078 commented on Jul 2, 2025 • 0 new comments

• Add `enable_parallel_tool_execution` flag to LlmAgent

  #1076 commented on Jul 2, 2025 • 0 new comments

• Fetching Image as a Tool for the agent

  #681 commented on Jul 2, 2025 • 0 new comments

• RestApiTool requires tool_context for authentication

  #510 commented on Jul 2, 2025 • 0 new comments

• Is ADK also planning to support "think" tool in the upcoming feature?

  #399 commented on Jul 2, 2025 • 0 new comments

• Enhance: Support Service Account Application Default Credentials on `google.adk.tools.google_api_tool.google_api_tool_sets` module

  #80 commented on Jul 2, 2025 • 0 new comments

• MCP tool execution is too slow

  #1643 commented on Jul 2, 2025 • 0 new comments

• Bug: Failed to detach context

  #423 commented on Jul 2, 2025 • 0 new comments

• Is there a way to get response to image input?

  #1350 commented on Jul 2, 2025 • 0 new comments

• Live Api with tools agents end with `extra_headers` issue

  #1212 commented on Jul 2, 2025 • 0 new comments

• Model does not stop output when new content is received.

  #1187 commented on Jul 2, 2025 • 0 new comments

• Problem with handling a WS connection break between the server and the LLM

  #1105 commented on Jul 2, 2025 • 0 new comments

• trying runner.run() using fastapi

  #1101 commented on Jul 2, 2025 • 0 new comments

• Restoring a saved session from the database for a Bidi streaming Agent doesn't work

  #1317 commented on Jul 2, 2025 • 0 new comments

• Share data between agents in liveAPI

  #1252 commented on Jul 2, 2025 • 0 new comments

• GCP agent engine support to deploy live voice agents using the gemini live-001 bi-directional voice model

  #1469 commented on Jul 2, 2025 • 0 new comments

• Enable ADK Agents to Call External Agents (e.g., OpenAI Agents SDK) as Tools

  #382 commented on Jul 2, 2025 • 0 new comments

• Structured Output (output schema) + Tool Call

  #701 commented on Jul 2, 2025 • 0 new comments

• Unable to deploy adk-python demo weather agent to Vertex AI

  #1004 commented on Jul 2, 2025 • 0 new comments

• I want to get "real" stream response from Agent Engine

  #1510 commented on Jul 3, 2025 • 0 new comments

• Sub-agents flow is interrupted in the bidi streaming context

  #1174 commented on Jul 3, 2025 • 0 new comments

• Cannot combine LoopAgent, state injection, and curly braces due to lack of global_instructions support

  #1235 commented on Jul 3, 2025 • 0 new comments

• ADK: Missing Thought Process Output with Claude Sonnet 4 Despite Enabled Parameters

  #1489 commented on Jul 3, 2025 • 0 new comments

• how get deepseek-r1 response reasoning_content

  #346 commented on Jul 3, 2025 • 0 new comments

• Using agent as tool outside of adk web doesn't exit cleanly

  #1112 commented on Jul 3, 2025 • 0 new comments

• OpenAI with LiteLLM tool usage throws error that doesnt appear when using Gemini or Claude

  #1389 commented on Jul 3, 2025 • 0 new comments

• before_model_callback not being called during run_live

  #1108 commented on Jul 3, 2025 • 0 new comments

• opentelemetry ValueError: "Token was created in a different Context" with ParallelAgent and LlmAgent

  #860 commented on Jul 3, 2025 • 0 new comments

• Conversation Role Alternation Error with Gemma-3-12b-it Model: litellm.BadRequestError

  #1373 commented on Jul 4, 2025 • 0 new comments

• Random "MALFORMED_FUNCTION_CALL" Error from Gemini Agent

  #1521 commented on Jul 4, 2025 • 0 new comments

• Ways to reduce context or event history sent to orchestrator llm

  #1032 commented on Jul 4, 2025 • 0 new comments

• Using proxy for AzureOpenAI: variable HTTPS_PROXY not honored

  #1650 commented on Jul 4, 2025 • 0 new comments

• Add Built-in Retry Mechanism for API Errors (e.g., 429 Too Many Requests) in LLM Agent

  #1214 commented on Jul 4, 2025 • 0 new comments

• gemini-live-2.5-flash-preview doesn't call tools

  #1698 commented on Jul 4, 2025 • 0 new comments

• LiteLLM giving error with OpenAI models and Grafana's MCP server

  #929 commented on Jul 4, 2025 • 0 new comments

• support PDF file in Lite_llm

  #363 commented on Jul 4, 2025 • 0 new comments

• agent_engines: TypeError: cannot pickle '_contextvars.Context' object

  #856 commented on Jul 4, 2025 • 0 new comments

• fix(litellm): prevent 400 error during parallel tool calls

  #231 commented on Jul 3, 2025 • 0 new comments

• feat: Added mem0 memory service

  #264 commented on Jul 1, 2025 • 0 new comments

• feat: ✨ redis session class

  #789 commented on Jul 2, 2025 • 0 new comments

• feat: Add BM25 retriever for in-memory memory service - issue #786

  #872 commented on Jul 4, 2025 • 0 new comments

• feat: enable access session from ReadonlyContext

  #1052 commented on Jul 1, 2025 • 0 new comments

• Adds support for custom vector stores using Langchain.

  #1072 commented on Jul 1, 2025 • 0 new comments

• feat: Add support for vertex gemini model optimizer

  #1130 commented on Jun 29, 2025 • 0 new comments

• feat(tools): request answer tool factory

  #1150 commented on Jul 1, 2025 • 0 new comments

• feat(tool): tool_error_notifier

  #1177 commented on Jul 1, 2025 • 0 new comments

• fix: Notify agent of failed agent transfer in function response

  #1206 commented on Jun 29, 2025 • 0 new comments

• chore: Split ADK Web into endpoint definitions and implementations

  #1217 commented on Jul 2, 2025 • 0 new comments

• Fix: Handle Pydantic v2 ValidationError in AgentLoader

  #1245 commented on Jul 1, 2025 • 0 new comments

• feat(sessions): Efficient Large Context Handling for Agent Development Kit

  #1247 commented on Jun 29, 2025 • 0 new comments

• Fix ollama provider error.

  #1344 commented on Jun 28, 2025 • 0 new comments

• feat: add optional per-agent Vertex AI project and location configuration

  #1431 commented on Jun 29, 2025 • 0 new comments

• feat: Introduce Retry on Netwok Error on Load Web Page

  #1433 commented on Jul 1, 2025 • 0 new comments

• fix: make LiteLLM streaming truly asynchronous

  #1451 commented on Jun 29, 2025 • 0 new comments

• docs: update StreamableHTTPConnectionParams docstring to remove SSE reference

  #1456 commented on Jun 27, 2025 • 0 new comments

• feat: Pass Auth through from AgentTool

  #1462 commented on Jun 29, 2025 • 0 new comments

• feat: Implement auto rater-based evaluator for responses

  #1499 commented on Jul 2, 2025 • 0 new comments

• fix(agent_graph): Prevent duplicate edges in agent graph (#1601)

  #1602 commented on Jul 4, 2025 • 0 new comments

• feat: cli funcionality to deploy an Agent to a running GKE cluster

  #1607 commented on Jul 3, 2025 • 0 new comments

• fix: Starred tuple unpacking everywhere in GCS artifact service

  #1669 commented on Jul 4, 2025 • 0 new comments