Issues

• LlmAgent(mode="task") with a raw dict output_schema fails at construction with SchemaError: Unknown schema type: "object"

  google/adk-python#7935 · Filter by author yifanxiong272 opened on Nov 2, 2026
  1

  

• Runner.run_async runs an LlmAgent with neither new_message nor invocation_id instead of raising ValueError

  google/adk-python#7934 · Filter by author yifanxiong272 opened on Nov 2, 2026
  1

  

• Workflow parent retry loses a deferred route when the node also writes deltas

  core
  Bug

  google/adk-python#7930 · Filter by author 1828703450qqcom-ux opened on Nov 2, 2026
  

• LlmAgent output_schema silently becomes an empty Schema() for Literal and Annotated types

  google/adk-python#7927 · Filter by author sumedh-aerram opened on Nov 1, 2026
  

• thinking_config is silently dropped with use_interactions_api=True

  agent config
  Bug

  google/adk-python#7924 · Filter by author jhackshaw opened on Nov 1, 2026
  

• RemoteA2aAgent in task mode hangs Runner.run_async() when the caller stops at its error event

  google/adk-python#7918 · Filter by author hyeonsang010716 opened on Nov 1, 2026
  

• Non-Gemini tool schemas reject None for nullable parameters

  tools
  Bug

  google/adk-python#7916 · Filter by author sumedh-aerram opened on Nov 1, 2026
  

• FileArtifactService changes CRLF and CR line endings in text artifacts

  services
  Bug

  google/adk-python#7914 · Filter by author sumedh-aerram opened on Oct 31, 2026
  

• ContainerCodeExecutor drops output files that are hard links

  google/adk-python#7906 · Filter by author AkshayShah03 opened on Oct 31, 2026
  2

  

• Resuming a workflow re-runs the earlier no-output runs of a wait_for_output node, so its side effects happen twice

  agent engine
  Bug

  google/adk-python#7905 · Filter by author AkshayShah03 opened on Oct 31, 2026
  

• LangchainTool.from_config blanks the name and description a config leaves out

  tools
  Bug

  google/adk-python#7900 · Filter by author roydonsequeira opened on Oct 31, 2026
  

• LlmAgent.clone() keeps the original's sub-agent tools and finish_task

  google/adk-python#7894 · Filter by author hyeonsang010716 opened on Oct 31, 2026
  

• Telemetry drops the inference span of a sub-agent that inherits a non-Gemini model

  tracing
  Bug

  google/adk-python#7892 · Filter by author adityasingh2400 opened on Oct 31, 2026
  

• DebugLoggingPlugin writes nothing for a failed run and keeps its state in memory

  core
  Bug

  google/adk-python#7890 · Filter by author adityasingh2400 opened on Oct 31, 2026
  

• Multi-turn trajectory and hallucinations metrics fail on a tool value with a datetime, Decimal or UUID

  google/adk-python#7888 · Filter by author adityasingh2400 opened on Oct 31, 2026
  

• PerTurnUserSimulatorQualityV1 scores unjudged turns as failures

  eval
  Bug

  google/adk-python#7886 · Filter by author adityasingh2400 opened on Oct 31, 2026
  

• Interactions API drops media a tool returns (FunctionResponse.parts)

  tools
  Bug

  google/adk-python#7884 · Filter by author adityasingh2400 opened on Oct 31, 2026
  

• Failed Interactions API calls always report UNKNOWN_ERROR / "Unknown error"

  google/adk-python#7882 · Filter by author adityasingh2400 opened on Oct 31, 2026
  

• CLI unit tests run the real gcloud: an ADC browser login on every run, and they unset the default project

  tools
  Bug

  google/adk-python#7880 · Filter by author adityasingh2400 opened on Oct 31, 2026
  

• BuiltInCodeExecutor emits an empty final event before every model reply

  agent engine
  Bug

  google/adk-python#7878 · Filter by author adityasingh2400 opened on Oct 31, 2026
  

• AgentDetails.tool_declarations come back as dicts after a JSON round trip, breaking judge metrics on saved runs

  google/adk-python#7876 · Filter by author adityasingh2400 opened on Oct 31, 2026
  

• LocalEvalService.evaluate adds rubrics to the caller's inference results, so scoring them again raises

  eval
  Bug

  google/adk-python#7874 · Filter by author adityasingh2400 opened on Oct 31, 2026
  

• ReflectAndRetryModelPlugin counts one streamed model error twice

  models
  Bug

  google/adk-python#7872 · Filter by author adityasingh2400 opened on Oct 31, 2026
  

• output_schema is silently dropped when Gemini uses the Interactions API (use_interactions_api=True)

  google/adk-python#7870 · Filter by author himanshu096 opened on Oct 30, 2026
  1

  

• BigQuery Agent Analytics plugin: long text lost when gcs_bucket_name is set with log_multi_modal_content=False, and inline binary parts logged as text/plain when not uploaded

  google/adk-python#7868 · Filter by author karolpiotrowicz opened on Oct 30, 2026
  4