Google ADK 2.14.0 Graduates Core Modules, Adds Teams Runner

Google shipped Agent Development Kit 2.14.0 on November 7, graduating execution environments and A2A out of experimental and wiring ADK agents into Microsoft Teams. The release summarizes itself on GitHub as one that "graduates execution environments, A2A, evaluation, and core runtime modules from experimental, adds multi-run evaluation and semantic tool argument matching, adds Microsoft Teams and OpenAI Decisions integrations, and makes long multi-turn workflows faster."

The headline additions. A new `TeamsRunner` connects ADK agents to Microsoft Teams, `MultiRunConfig` runs each eval case multiple times concurrently, and `ToolArgsMatcher` compares tool arguments using "exact, case-insensitive, numeric-tolerance, or LLM-backed semantic checks." `MongoDbMemoryService` adds vector-search retrieval with TTL expiration, and `adk generate_agent_card` builds A2A agent cards from the CLI.

Google's performance claim is specific: "a 30-node workflow's 200-turn run is 23% faster with 9% lower peak memory." The release notes publish no baseline hardware, no second benchmark, and no comparison against a named prior ADK version.

Three breaking changes ride along. Enum tool arguments now arrive as enum members rather than raw strings; `VertexAiSessionService` raises `ValueError` on extra keywords; and `RunSkillScriptTool` no longer accepts a free-form `command` parameter, restricting runs to validated `.py`, `.sh`, and `.bash` scripts from a skill directory. Twenty-six auth, environment, app, CLI and cache modules move to stable alongside 16 toolset features, making this Google's 210th tracked story in our Google feed over the last 90 days.