[Release] google/adk-python v1.26.0: v1.26.0

Google ADK Python v1.26.0 release introduces major features including intra-invocation token compaction, Apigee LLM chat completions integration, Agent Skills spec compliance, BigQuery search catalog support, and an Agent Registry system. The release also includes numerous bug fixes for streaming tools, memory services, authentication, and various integrations across core, models, tools, evals, and services components.

Key Points

• •Implement intra-invocation and token compaction pre-request to optimize memory and token usage during agent invocations
• •Integrate `/chat/completions` endpoint with ApigeeLlm including streaming support for improved LLM compatibility
• •Add `load_skill_from_dir()` method and ensure Agent Skills spec compliance with validation, aliases, scripts, and auto-injection
• •Introduce User Personas to the evaluation framework for more realistic agent testing and persona-based behavior validation
• •Enable memory consolidation via Vertex AI Memory Bank with generate/create modes for persistent agent context management
• •Implement interceptor framework in `A2aAgentExecutor` for flexible agent-to-agent communication control and extensibility
• •Add Agent Registry integration to discover, manage, and create AgentCards from agent endpoints with schema auto-upgrade
• •Support native ID token handling in OAuth2 credentials and `ServiceAccountCredentialExchanger` for enhanced authentication
• •Parallelize tool resolution in `LlmAgent.canonical_tools()` and fix streaming tool detection for improved performance
• •Enable dependency injection for agent loader in FastAPI app generation for flexible deployment configurations

Workflow Diagram

Concepts