The ADK Python repository has been updated from v1.18.0 to v1.19.0. This release includes several new features, API changes, and bug fixes that need to be reflected in the documentation.
-
Update Python Version Requirement.
The minimum required Python version has been updated from 3.9 to 3.10.
Current state:
The documentation in docs/index.md and other places might still mention Python 3.9.
Proposed Change:
Update all mentions of the Python version requirement to 3.10.
Reasoning:
The pyproject.toml file in the ADK Python repository has been updated to require Python 3.10 or higher. This change needs to be reflected in the documentation to ensure users have the correct environment setup.
Reference:
-
Document the New Service Registration Feature.
A new service registry has been introduced that allows users to register custom services for sessions, artifacts, and memory using a services.py or services.yaml file.
Current state:
There is no documentation for this new feature.
Proposed Change:
Add a new section to the documentation that explains how to register custom services. This section should include:
- An explanation of the service registry and its purpose.
- Instructions on how to create and use
services.yaml and services.py.
- Code examples for both methods of registration.
Reasoning:
This is a new feature that provides a powerful way for users to extend the ADK with their own custom services. The documentation should be updated to make users aware of this feature and how to use it.
Reference:
src/google/adk/cli/service_registry.py
contributing/samples/services.py
contributing/samples/services.yaml
-
Document the New FileArtifactService and Artifact Service Updates.
A new FileArtifactService has been added for storing artifacts on the local filesystem. Additionally, the save_artifact method in CallbackContext now accepts custom_metadata, and new methods get_artifact_version and list_artifacts have been added.
Current state:
The documentation for artifact services may not include information about the FileArtifactService or the recent updates to CallbackContext.
Proposed Change:
- Add documentation for the
FileArtifactService, explaining how to configure and use it.
- Update the documentation for
CallbackContext to include the new custom_metadata parameter for save_artifact and the new get_artifact_version and list_artifacts methods.
- Add a new sample agent that demonstrates the use of the
FileArtifactService and the new artifact-related methods.
Reasoning:
These updates provide more flexibility for managing artifacts. The documentation should be updated to reflect these new capabilities.
Reference:
src/google/adk/artifacts/file_artifact_service.py
src/google/adk/agents/callback_context.py
contributing/samples/context_offloading_with_artifact/
-
Document the Session Rewind Feature.
A new feature has been added that allows rewinding a session to a previous state.
Current state:
There is no documentation for this new feature.
Proposed Change:
Add a new section to the documentation that explains the session rewind feature, how to use it, and provides a code example.
Reasoning:
This is a useful feature for debugging and replaying sessions. The documentation should be updated to make users aware of this feature.
Reference:
contributing/samples/rewind_session/main.py
src/google/adk/runners.py (the rewind_async method)
-
Document the on_model_error_callback.
A new on_model_error_callback has been added to the LlmAgent class to allow for custom error handling when a model call fails.
Current state:
There is no documentation for this new callback.
Proposed Change:
Update the documentation for LlmAgent to include the new on_model_error_callback, explaining its purpose, how to use it, and providing a code example.
Reasoning:
This new callback provides a way for users to implement custom error handling and recovery logic. The documentation should be updated to reflect this new capability.
Reference:
src/google/adk/agents/llm_agent.py
-
Update Documentation for save_live_blob.
The save_live_audio run configuration has been deprecated and replaced with save_live_blob.
Current state:
The documentation may still refer to save_live_audio.
Proposed Change:
Update all mentions of save_live_audio to save_live_blob and explain that the new configuration supports both audio and video data.
Reasoning:
This is a breaking change for users of the live/bidi-streaming features. The documentation needs to be updated to reflect this change.
Reference:
src/google/adk/agents/run_config.py
-
Document New and Removed Plugins.
The BigQueryAgentAnalyticsPlugin and MultimodalToolResultsPlugin have been added, and the bigquery_logging_plugin has been removed.
Current state:
The documentation for plugins may be out of date.
Proposed Change:
- Add documentation for the
BigQueryAgentAnalyticsPlugin and MultimodalToolResultsPlugin, explaining their purpose, how to configure them, and how to use them.
- Remove documentation for the
bigquery_logging_plugin.
Reasoning:
These changes to the available plugins need to be reflected in the documentation.
Reference:
src/google/adk/plugins/bigquery_agent_analytics_plugin.py
src/google/adk/plugins/multimodal_tool_results_plugin.py
-
Document Session Service Updates.
A new SqliteSessionService has been added, and the DatabaseSessionService is now async.
Current state:
The documentation for session services may not include these updates.
Proposed Change:
- Add documentation for the
SqliteSessionService.
- Update the documentation for
DatabaseSessionService to reflect its new async nature.
Reasoning:
These are significant changes to the session management capabilities of the ADK. The documentation should be updated to reflect these changes.
Reference:
src/google/adk/sessions/sqlite_session_service.py
src/google/adk/sessions/database_session_service.py
-
Add Documentation for New Sample Agents.
Several new sample agents have been added to demonstrate new features.
Current state:
There is no documentation for these new sample agents.
Proposed Change:
Add documentation for the following new sample agents:
context_offloading_with_artifact
litellm_inline_tool_call
litellm_structured_output
mcp_in_agent_tool_remote
mcp_in_agent_tool_stdio
multimodal_tool_results
Reasoning:
These new samples demonstrate important new features of the ADK. The documentation should be updated to make users aware of these samples and the features they demonstrate.
Reference:
The ADK Python repository has been updated from v1.18.0 to v1.19.0. This release includes several new features, API changes, and bug fixes that need to be reflected in the documentation.
Compare Link: google/adk-python@v1.18.0...v1.19.0
Here are the recommended documentation updates:
Update Python Version Requirement.
The minimum required Python version has been updated from 3.9 to 3.10.
Current state:
The documentation in
docs/index.mdand other places might still mention Python 3.9.Proposed Change:
Update all mentions of the Python version requirement to 3.10.
Reasoning:
The
pyproject.tomlfile in the ADK Python repository has been updated to require Python 3.10 or higher. This change needs to be reflected in the documentation to ensure users have the correct environment setup.Reference:
pyproject.tomlDocument the New Service Registration Feature.
A new service registry has been introduced that allows users to register custom services for sessions, artifacts, and memory using a
services.pyorservices.yamlfile.Current state:
There is no documentation for this new feature.
Proposed Change:
Add a new section to the documentation that explains how to register custom services. This section should include:
services.yamlandservices.py.Reasoning:
This is a new feature that provides a powerful way for users to extend the ADK with their own custom services. The documentation should be updated to make users aware of this feature and how to use it.
Reference:
src/google/adk/cli/service_registry.pycontributing/samples/services.pycontributing/samples/services.yamlDocument the New
FileArtifactServiceand Artifact Service Updates.A new
FileArtifactServicehas been added for storing artifacts on the local filesystem. Additionally, thesave_artifactmethod inCallbackContextnow acceptscustom_metadata, and new methodsget_artifact_versionandlist_artifactshave been added.Current state:
The documentation for artifact services may not include information about the
FileArtifactServiceor the recent updates toCallbackContext.Proposed Change:
FileArtifactService, explaining how to configure and use it.CallbackContextto include the newcustom_metadataparameter forsave_artifactand the newget_artifact_versionandlist_artifactsmethods.FileArtifactServiceand the new artifact-related methods.Reasoning:
These updates provide more flexibility for managing artifacts. The documentation should be updated to reflect these new capabilities.
Reference:
src/google/adk/artifacts/file_artifact_service.pysrc/google/adk/agents/callback_context.pycontributing/samples/context_offloading_with_artifact/Document the Session Rewind Feature.
A new feature has been added that allows rewinding a session to a previous state.
Current state:
There is no documentation for this new feature.
Proposed Change:
Add a new section to the documentation that explains the session rewind feature, how to use it, and provides a code example.
Reasoning:
This is a useful feature for debugging and replaying sessions. The documentation should be updated to make users aware of this feature.
Reference:
contributing/samples/rewind_session/main.pysrc/google/adk/runners.py(therewind_asyncmethod)Document the
on_model_error_callback.A new
on_model_error_callbackhas been added to theLlmAgentclass to allow for custom error handling when a model call fails.Current state:
There is no documentation for this new callback.
Proposed Change:
Update the documentation for
LlmAgentto include the newon_model_error_callback, explaining its purpose, how to use it, and providing a code example.Reasoning:
This new callback provides a way for users to implement custom error handling and recovery logic. The documentation should be updated to reflect this new capability.
Reference:
src/google/adk/agents/llm_agent.pyUpdate Documentation for
save_live_blob.The
save_live_audiorun configuration has been deprecated and replaced withsave_live_blob.Current state:
The documentation may still refer to
save_live_audio.Proposed Change:
Update all mentions of
save_live_audiotosave_live_bloband explain that the new configuration supports both audio and video data.Reasoning:
This is a breaking change for users of the live/bidi-streaming features. The documentation needs to be updated to reflect this change.
Reference:
src/google/adk/agents/run_config.pyDocument New and Removed Plugins.
The
BigQueryAgentAnalyticsPluginandMultimodalToolResultsPluginhave been added, and thebigquery_logging_pluginhas been removed.Current state:
The documentation for plugins may be out of date.
Proposed Change:
BigQueryAgentAnalyticsPluginandMultimodalToolResultsPlugin, explaining their purpose, how to configure them, and how to use them.bigquery_logging_plugin.Reasoning:
These changes to the available plugins need to be reflected in the documentation.
Reference:
src/google/adk/plugins/bigquery_agent_analytics_plugin.pysrc/google/adk/plugins/multimodal_tool_results_plugin.pyDocument Session Service Updates.
A new
SqliteSessionServicehas been added, and theDatabaseSessionServiceis now async.Current state:
The documentation for session services may not include these updates.
Proposed Change:
SqliteSessionService.DatabaseSessionServiceto reflect its new async nature.Reasoning:
These are significant changes to the session management capabilities of the ADK. The documentation should be updated to reflect these changes.
Reference:
src/google/adk/sessions/sqlite_session_service.pysrc/google/adk/sessions/database_session_service.pyAdd Documentation for New Sample Agents.
Several new sample agents have been added to demonstrate new features.
Current state:
There is no documentation for these new sample agents.
Proposed Change:
Add documentation for the following new sample agents:
context_offloading_with_artifactlitellm_inline_tool_calllitellm_structured_outputmcp_in_agent_tool_remotemcp_in_agent_tool_stdiomultimodal_tool_resultsReasoning:
These new samples demonstrate important new features of the ADK. The documentation should be updated to make users aware of these samples and the features they demonstrate.
Reference:
contributing/samples/