Skip to content
Agents tracked: 258 Downloads (7d): 219M up 6.1% GitHub stars: 5.5M VS Code installs: 148M Releases (7d): 293 Agent status: 2 with issues Updated Oct 7, 2026

Spring AI Alibaba changelog: what's new each month

Every stable Spring AI Alibaba release summarised by month: the highlights, new features, improvements, fixes and anything you need to act on. 3 months covered; the current month updates daily.

March 2026

2 releases: 1.1.2.1 → 1.1.2.2

March brought significant enhancements to agent orchestration and workflow capabilities in Spring AI Alibaba. The release introduced multiple patterns for coordinating multiple agents and managing complex task delegation.

Highlights

  • AgentScope integration wraps ReActAgent for use in graph workflows.
  • Subagent pattern enables task delegation to specialized agents via Task/TaskOutput tools.
  • Supervisor pattern centralizes control over multiple tool-based agents like calendar and email.
  • Skills system allows agents to load skill content on demand with progressive disclosure.
  • Routing patterns (simple and graph-based) enable intelligent query classification and parallel agent invocation.

New

  • AgentScopeAgent wraps AgentScope ReActAgent as a BaseAgent for graph workflows.
  • Subagent orchestration with task delegation to specialized sub-agents (codebase explorer, web researcher, etc.).
  • Support for both Markdown and API-defined sub-agents in subagent pattern.
  • Supervisor agent pattern that wraps calendar and email agents as tools and invokes them on demand.
  • Skills loading system with on-demand skill content retrieval and progressive disclosure of skill descriptions.
  • Simple LlmRoutingAgent for query classification and parallel specialist agent invocation (GitHub, Notion, Slack).
  • Graph-based routing with LlmRoutingAgent as StateGraph node including preprocess, postprocess, and merge nodes.
  • Handoff patterns supporting single-agent step progression and multi-agent conditional routing between agents.

Improved

  • ModelInterceptor now injects step-specific system prompt and tools per turn for single-agent handoffs.
  • System prompt optimization shows only skill descriptions instead of full skill content for smaller context windows.

Fixed

  • Several bugs in version 1.1.2.1 were resolved in version 1.1.2.2.

February 2026

1 release: 1.1.2.0

February brought significant enhancements to agent capabilities in Spring AI Alibaba, with new multi-agent patterns, improved streaming APIs, and better tool execution control.

Highlights

  • Agent skills now integrated into ReactAgent with progressive disclosure support.
  • Built-in multi-agent patterns for routing, supervisor, handoffs, and skills with parallel execution.
  • New streamMessages API simplifies message streaming in agent applications.
  • Tools can now execute in parallel and asynchronously for better performance.
  • Flow agent hooks allow customization across all agent types.

New

  • Agent skills capabilities in ReactAgent
  • Multi-agent pattern support including routing, supervisor, handoffs, and skills
  • Parallel execution for routing and supervisor patterns
  • Flow agent hook support for all flow agent types
  • streamMessages API for simplified message streaming
  • New overload invoke and call methods accepting extra inputs
  • Tool returnDirect behavior
  • Parallel conditional edges execution

Improved

  • Tool execution now supports parallelization and asynchronous execution
  • ToolContextHelper class added for easier metadata access in tool context

Fixed

  • Tool context metadata access issue

December 2025

1 release: 1.1.0.0

Spring AI Alibaba 1.1.0.0 was released as the official version, built on Spring AI 1.1.0 and Spring AI Alibaba Extensions 1.1.0.0. Documentation and upgrade guidance are available on the project website.

Summaries are written automatically from the official release notes (full changelog ↗); check the original notes before relying on a detail. Spring AI Alibaba: pricing, features and alternatives · All changelogs