v0.2.0
AI collaboration: guardrails for coding agents in generated projects, actionable architecture reports, and use cases as MCP tools.
Added
- Generated projects ship
AGENTS.mdandCLAUDE.mddescribing layers, conventions and themvn verifygate for AI coding agents - Claude Code Skills in generated projects:
ddk-add-aggregate,ddk-add-use-caseandddk-add-domain-eventfor the four-layer template,ddk-add-featurefor the three-layer template - Generated projects include
spring-boot-starter-webmvc-testfor MockMvc tests AGENTS.mdandCLAUDE.mdfor contributors to DDK itselfddk-mcp-starter: exposes application use cases as MCP tools on top of Spring AI 2.0, with streamable HTTP by default, Bean Validation on tool arguments, error-coded tool errors, hidden internals for unexpected exceptions and an audit log line per callCommonArchRules.MCP_TOOLS_MUST_RESIDE_IN_ADAPTERkeeps@McpToolmethods in the adapter layer- The user example exposes
register_user,get_useranddisable_useras MCP tools ArchGuard.checkevaluates all architecture rules at once and writestarget/archguard/violations.jsonandviolations.md, each violation with its rule, class and a concrete fix; generated projects and the example use it
Changed
- Generated projects and the example check architecture in one
ArchitectureTest.architectureIsRespected()throughArchGuard.check, instead of one test per rule - The planned
ddk-ai-starteris dropped: chat client configuration, structured output and token usage metrics are built into Spring AI 2.0
Fixed
scripts/ddk.shinstallsX.Y.Zinstead of the snapshot version whenDDK_REFnames a release tag such asv0.1.0