chore(main): release 0.6.0

[tilgalas]

🤖 I have created a release beep boop

0.6.0 (2026-02-19)

Features

• Add Compact processor to SingleFlow (ee459b3)
• Add Compaction RequestProcessor for event compaction in llm flow (af1fafe)
• Add ContextCacheConfig to InvocationContext (968a9a8)
• Add event compaction config to InvocationContext (8f7d7ea)
• Add event compaction framework in Java ADK (dd68c85)
• add eventId in CallbackContext and ToolContext (ac05fde)
• add ExampleTool to ComponentRegistry (2e1b09f)
• add response converters to support multiple A2A client events (4e8de90)
• Add token usage threshold to TailRetentionEventCompactor (9901307)
• Add tokenThreshold and eventRetentionSize to EventsCompactionConfig (588b00b)
• Add VertexAiSearchTool and AgentTools for search (b48b194)
• Adding a .close() method to Runner, Agent and Plugins (495bf95)
• Adding a new ArtifactService.saveAndReloadArtifact() method (59e87d3)
• adding a new temporary store of context for callbacks (ed736cd)
• Adding autoCreateSession in Runner (6dd51cc)
• Adding GlobalInstructionPlugin (72e20b6)
• Adding OnModelErrorCallback (dfd2944)
• adding resume / event management primitives (2de03a8)
• Adding TODO files for reaching idiomatic java (4ac1dd2)
• Adding validation to BaseAgent (5dfc000)
• Adding validation to BaseAgent and RunConfig (503caa6)
• Adding validation to InvocationContext 'session_service', 'invocation_id', (0502c21)
• Allow EventsCompactionConfig to have a null summarizer initially (229654e)
• enable LoopAgent configuration (d1a1cea)
• EventAction.stateDelta() now has a remove by key variant (32a6b62)
• Extend google_search support to Gemini 3 in Java ADK (ddb00ef)
• Fix a handful of small changes related to headers, logging and javadoc (0b63ca3)
• Forward state delta to parent session (00d6d30)
• HITL - remove the events between the confirmed FC & its response (3670555)
• HITL - Revert the "Boolean confirmation" changes, we'll fix it differently (f65e58b)
• HITL: Declining a proposal now correctly intercepts the run (9611f89)
• HITL: Let ADK resume after HITL approval is present (9611f89)
• Improving LoggingPlugin (acfaa04)
• Integrate event compaction in Java ADK runner (54c826c)
• Introduce TailRetentionEventCompactor to compact and retain the tail of the event stream (efe58d6)
• Introduce the App class for defining agentic applications (d7c5c6f)
• introduces context caching configuration for apps, ported from Python ADK (12defee)
• new ContextFilterPlugin (f8e9bc3)
• Refactor EventsCompactionConfig to require a summarizer (864d606)
• refactor remote A2A agent to use A2A SDK client (7792233)
• Refine bug and feature request issue templates (3e74c9a)
• register GoogleMapsTool in ComponentRegistry (464f0b2)
• Reorder compaction events in chronological order (66e2296)
• Setting up data structures for pause/resume/rewind (c6c52c4)
• Skip post-invocation compaction if parameters not set (76f86c5)
• Support function calls in LLM event summarizer (55144ac)
• support stdio_connection_params in McpToolset config (cc1588a)
• Token count estimation fallback for tail retention compaction (3338565)
• Update event compaction logic to include events after compaction end times (ea12505)
• Updating Baseline Code executors (a3f1763)
• updating Telemetry (5ba63f4)
• Updating the Tracing implementation and updating BaseAgent.runLive (8acb1ea)
• use Credentials' request metadata to populate headers (e01df11)

Bug Fixes

• Add name and description to configagent pom.xml (4948bfc)
• Align InMemorySessionService listSessions with Python implementation (9434949)
• Always use a mutable HashMap for default function arguments (c6c9557)
• emit multiple LlmResponses in GeminiLlmConnection (7bf55f1)
• Events for HITL are now emitted correctly (9611f89)
• fix linter error (f49260e)
• Fixing a problem with serializing sessions that broke integration with Vertex AI Session Service (8190ed3)
• Fixing a regression in InMemorySessionService (d11bedf)
• Fixing Vertex session storage (5607f64)
• HITL endless loop when asking for approvals (9611f89)
• include usage_metadata events in live postprocessing (8137d66)
• javadocs in ResponseConverter (be35b22)
• Make FunctionResponses respect the order of FunctionCalls (a99c75b)
• Making stepsCompleted thread-safe (d432c64)
• Merging of events in rearrangeEventsForAsyncFunctionResponsesInHistory (67c29e3)
• Mutate EventActions in-place in AgentTool (ded5a4e)
• pass mutable function args map to beforeToolCallback (e989ae1)
• populate finishReason in LlmResponse (dace210)
• Propagate trace context across async boundaries (279c977)
• recursively extract input/output schema for AgentTool (7019d39)
• Reduce the logging level (dd601ca)
• Remove checking ToolConfirmation from Functions to align with Python SDK (0724330)
• remove client-side function call IDs from LlmRequest (99b5fc2)
• Remove obsolete @param tags from SessionController Javadoc (a77971a)
• Replace @apiNote with <p> in Javadoc comments. (ac16d53)
• restore invocationContext() method (c9e2a5b)
• revert: Merging of events in rearrangeEventsForAsyncFunctionResponsesInHistory (101adce)
• update converters package classes (b66e4a5)
• update EmbeddingModelDiscoveryTest package statement (adeb9dc)
• Updated BasePlugin JavaDoc for name parameter (2e59550)

Documentation

• Update comment in Runner (fe00ef8)

---

This PR was generated with Release Please. See documentation.

[gemini-code-assist]

Summary of Changes

Hello @tilgalas, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This release delivers a substantial upgrade to the Java ADK, primarily by introducing an advanced event compaction framework to optimize event processing and memory usage. It also significantly refines the Human-in-the-Loop (HITL) features, making agent interactions more robust and controllable. Beyond these major additions, the update integrates new tools and plugins, enhances core functionalities with better validation and resource management, and resolves numerous issues to boost the stability and reliability of the toolkit.

Highlights

• Major Version Update: The project has been updated to version 0.6.0, reflecting a significant set of new features and improvements.
• Event Compaction Framework: A comprehensive event compaction framework has been introduced, including processors, configurability in InvocationContext, and a TailRetentionEventCompactor for efficient event stream management.
• Enhanced Human-in-the-Loop (HITL) Capabilities: Several improvements have been made to HITL functionality, such as correct interception on declining proposals, resuming after approval, and accurate event emission.
• New Tools and Plugins: New components like VertexAiSearchTool, AgentTools for search, GlobalInstructionPlugin, and ContextFilterPlugin have been added, expanding the toolkit's capabilities.
• Core Functionality Improvements: Key enhancements include adding a .close() method to Runner, Agent, and Plugins, validation for BaseAgent and RunConfig, and support for multiple A2A client events.
• Numerous Bug Fixes: A wide array of bug fixes address issues such as session serialization, thread safety, function argument handling, and trace context propagation, improving overall stability.

🧠 New Feature in Public Preview: You can now enable Memory to help Gemini Code Assist learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. Click here to enable Memory in your admin console.

Changelog

• Bug Fixes
  • Add name and description to configagent pom.xml
  • Align InMemorySessionService listSessions with Python implementation
  • Always use a mutable HashMap for default function arguments
  • emit multiple LlmResponses in GeminiLlmConnection
  • Events for HITL are now emitted correctly
  • fix linter error
  • Fixing a problem with serializing sessions that broke integration with Vertex AI Session Service
  • Fixing a regression in InMemorySessionService
  • Fixing Vertex session storage
  • HITL endless loop when asking for approvals
  • include usage_metadata events in live postprocessing
  • javadocs in ResponseConverter
  • Make FunctionResponses respect the order of FunctionCalls
  • Making stepsCompleted thread-safe
  • Merging of events in rearrangeEventsForAsyncFunctionResponsesInHistory
  • Mutate EventActions in-place in AgentTool
  • pass mutable function args map to beforeToolCallback
  • populate finishReason in LlmResponse
  • Propagate trace context across async boundaries
  • recursively extract input/output schema for AgentTool
  • Reduce the logging level
  • Remove checking ToolConfirmation from Functions to align with Python SDK
  • remove client-side function call IDs from LlmRequest
  • Remove obsolete @param tags from SessionController Javadoc
  • Replace @APinote with

    in Javadoc comments.

  • restore invocationContext() method
  • revert: Merging of events in rearrangeEventsForAsyncFunctionResponsesInHistory
  • update converters package classes
  • update EmbeddingModelDiscoveryTest package statement
  • Updated BasePlugin JavaDoc for name parameter
• Documentation
  • Update comment in Runner
• Features
  • Add Compact processor to SingleFlow
  • Add Compaction RequestProcessor for event compaction in llm flow
  • Add ContextCacheConfig to InvocationContext
  • Add event compaction config to InvocationContext
  • Add event compaction framework in Java ADK
  • add eventId in CallbackContext and ToolContext
  • add ExampleTool to ComponentRegistry
  • add response converters to support multiple A2A client events
  • Add token usage threshold to TailRetentionEventCompactor
  • Add tokenThreshold and eventRetentionSize to EventsCompactionConfig
  • Add VertexAiSearchTool and AgentTools for search
  • Adding a .close() method to Runner, Agent and Plugins
  • Adding a new ArtifactService.saveAndReloadArtifact() method
  • adding a new temporary store of context for callbacks
  • Adding autoCreateSession in Runner
  • Adding GlobalInstructionPlugin
  • Adding OnModelErrorCallback
  • adding resume / event management primitives
  • Adding TODO files for reaching idiomatic java
  • Adding validation to BaseAgent
  • Adding validation to BaseAgent and RunConfig
  • Adding validation to InvocationContext 'session_service', 'invocation_id',
  • Allow EventsCompactionConfig to have a null summarizer initially
  • enable LoopAgent configuration
  • EventAction.stateDelta() now has a remove by key variant
  • Extend google_search support to Gemini 3 in Java ADK
  • Fix a handful of small changes related to headers, logging and javadoc
  • Forward state delta to parent session
  • HITL - remove the events between the confirmed FC & its response
  • HITL - Revert the "Boolean confirmation" changes, we'll fix it differently
  • HITL: Declining a proposal now correctly intercepts the run
  • HITL: Let ADK resume after HITL approval is present
  • Improving LoggingPlugin
  • Integrate event compaction in Java ADK runner
  • Introduce TailRetentionEventCompactor to compact and retain the tail of the event stream
  • Introduce the App class for defining agentic applications
  • introduces context caching configuration for apps, ported from Python ADK
  • new ContextFilterPlugin
  • Refactor EventsCompactionConfig to require a summarizer
  • refactor remote A2A agent to use A2A SDK client
  • Refine bug and feature request issue templates
  • register GoogleMapsTool in ComponentRegistry
  • Reorder compaction events in chronological order
  • Setting up data structures for pause/resume/rewind
  • Skip post-invocation compaction if parameters not set
  • Support function calls in LLM event summarizer
  • support stdio_connection_params in McpToolset config
  • Token count estimation fallback for tail retention compaction
  • Update event compaction logic to include events after compaction end times
  • Updating Baseline Code executors
  • updating Telemetry
  • Updating the Tracing implementation and updating BaseAgent.runLive
  • use Credentials' request metadata to populate headers

Activity

• This pull request was automatically generated by the 'release-please' bot to publish version 0.6.0 of the project.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature	Command	Description
Code Review	/gemini review	Performs a code review for the current pull request in its current state.
Pull Request Summary	/gemini summary	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help	/gemini help	Displays a list of available commands.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

Footnotes

1. Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

[gemini-code-assist]

Code Review

This pull request appears to be an automated release for version 0.6.0. The changes correctly update the version number across various pom.xml files, the README.md, and other configuration files. The CHANGELOG.md has also been updated with a comprehensive list of changes. I have one minor suggestion regarding the release date in the changelog.

[gemini-code-assist]

The release date for version 0.6.0 is set to 2026-02-19, which is in the future. This is likely an error and should be corrected to the actual release date. It seems there might be a recurring issue with date generation, as the date for version 0.3.0 is also set in the future (2025-09-17).

[github-actions]

🤖 Created releases:

• v0.6.0

🌻