Agent Development Kit (ADK) 2.10.0 released
A feature release adding skill lifecycle management, MongoDB vector search integration, and richer evaluation metrics for tracking agent performance.
What changed in 2.10.0
- New experimental skill lifecycle modes and active skill limits to control resource usage and tool persistence (enable with ADK_ENABLE_SKILL_LIFECYCLE=1).
- MongoDB toolset with vector and hybrid search capabilities for agent flows.
- Evaluation metrics now track agent duration, token consumption, and model call counts.
- OpenAI reasoning models get automated request parameter adaptation and reasoning token reporting.
- BigQuery protected write mode rejects statements that cannot be placed in the session dataset; temporary tables no longer survive restarts.
Check before you update
- OpenAIResponsesLlm ignores thinking_config; use OpenAIGenerateContentConfig.effort instead.
- AgentEvaluator.evaluate and evaluate_eval_set raise ValueError when zero eval cases are evaluated.
- Instruction templating no longer fills ${var} and \{var} patterns from state.
Summarised by AgentGid from Google's official release notes on GitHub, as reported there; read the original before relying on it. Price, status and release dates come from our daily data.
Agent Development Kit (ADK) at a glance
| What it is | Google's open-source, code-first toolkit for building, evaluating and deploying AI agents, with SDKs for Python, TypeScript, Go, Java and Kotlin. |
|---|---|
| Latest version | 2.15.0, Nov 19, 2026. This page is about an earlier release. |
| Release pace | 4 stable releases in the last 30 days; 2.10.0 came 7 days after the previous one |
| Price | Free (OSS). Agent Development Kit (ADK) is open source and free to use; you pay only for the models and infrastructure it runs on. (checked Oct 14, 2026) |
| Downloads, last 7 days | 3.2M up 6.4% |
| Gid Score | 56 #10 in Agent frameworks and SDKs |
Agent Development Kit (ADK) overview Monthly changelogAlternatives