Starting a new Java project is slow — even with a coding agent helping. You set up Maven modules, wire databases and tests, copy boilerplate from that one project that "mostly works," and then spend the first hour explaining every architectural convention to Claude Code, Cursor, Copilot, or Codex from scratch. Hours pass before you write a single line of business logic. Trabuco exists because your time — and your agent's context window — is better spent building features, not fighting configuration or re-teaching conventions.
Trabuco is a command-line tool — and a Claude Code plugin — that generates both halves of a modern Java codebase: a complete, production-ready multi-module Maven project and the AI context that teaches coding agents how to work in it. Run trabuco init (or inside Claude Code, type /trabuco:new-project and describe what you need in plain English), answer a few prompts (or pass flags for automation), and you get a fully wired Spring Boot codebase alongside task-specific prompts, quality specifications, per-agent rule files, and workflow hooks already configured for Claude Code, Codex, Cursor, and GitHub Copilot. No templates to download, no manual setup, and no session spent bootstrapping your agent's understanding of the project.
The generated code is production-grade by default. Spring Boot with Spring Data JDBC (no JPA surprises), Flyway migrations, Testcontainers for real integration tests, Resilience4j circuit breakers, Google Java Format enforced by Spotless, ArchUnit rules that fail the build on layer violations, correlation-ID tracing, Prometheus metrics, OpenAPI + Swagger UI, and a global exception handler with sanitized responses. PostgreSQL, MySQL, MongoDB, or Redis — all configured with Docker Compose. JobRunr for background jobs; Kafka, RabbitMQ, AWS SQS, or GCP Pub/Sub for event-driven processing. The modular layout — Model, SQLDatastore / NoSQLDatastore, Shared, API, Worker, EventConsumer — has clean compile-time boundaries so API physically cannot import Worker code. Every opinion is deliberate: keyset pagination, no foreign-key constraints, Immutables at module boundaries, constructor injection only, bulk-bounded writes.
Alongside the code, Trabuco lays down an AI collaboration layer that the major coding agents load natively. The .ai/prompts/ directory ships task-specific guides (add-entity, add-endpoint, add-service, add-event, add-job, add-tool) plus JAVA_CODE_QUALITY.md — an authoritative specification covering architecture boundaries, exception handling, datastore performance (bulk I/O, keyset drain loops, denormalization), and testing standards. Per-agent rule files — CLAUDE.md for Claude Code, AGENTS.md for Codex, .cursor/rules/java.mdc for Cursor, .github/instructions/java.instructions.md for Copilot — wire those conventions into each tool's native discovery. Claude also gets .claude/skills/ for commit, PR, and review workflows; Codex and Cursor get hooks; Copilot gets setup steps. Every architectural convention lives in two places: enforced by the generated code and explained to the agents that will extend it.
Trabuco goes further. The AI Agent module ships production scaffolding for building AI agents themselves on Spring AI: tool calling, LLM input/output guardrails, multi-agent orchestration, knowledge-base integration, MCP server endpoint, the A2A (Agent-to-Agent) protocol, streaming responses, webhook callbacks, scope-based authorization, rate limiting, and correlation-ID tracing — wired, testable, and ready from the first commit. Trabuco itself is also an MCP server: run trabuco mcp and your coding agents invoke scaffolding operations natively (init_project, add_module, design_system, suggest_architecture).
Migrating an existing Spring Boot project is supported via
trabuco migrate: an orchestrated 14-phase migration that transforms a Maven-based Spring Boot repo in place into a Trabuco-shaped multi-module project. Each phase is owned by a specialist that proposes structured changes you approve at a gate before they're committed; per-phase git tags provide atomic rollback. Start withtrabuco migrate assess /path/to/repo. Full guide:docs/migration-guide.md.
Install
# Recommended: use via npx (no install needed)
claude mcp add trabuco -- npx -y trabuco-mcp
# Or install the CLI globally
npm install -g trabuco-mcp # via npm
go install github.com/arianlopezc/Trabuco/cmd/trabuco@latest # via Go
curl -sSL https://raw.githubusercontent.com/arianlopezc/Trabuco/main/scripts/install.sh | bash # via shell
For per-agent MCP setup (Cursor, Copilot, Codex), offline tarballs, and binary fallback, see docs/manual.md#installation.
Quick start
# Interactive — Trabuco asks what you need
trabuco init
# Or all flags
trabuco init --name=myapp --group-id=com.company.myapp \
--modules=Model,SQLDatastore,Shared,API --database=postgresql
cd myapp
mvn clean install
cd API && mvn spring-boot:run
Your API runs at http://localhost:8080. For Worker, EventConsumer, AIAgent, and CI examples, see docs/manual.md#quick-start.
Claude Code plugin
Trabuco is published in Anthropic's community plugin marketplace. Inside Claude Code:
/plugin marketplace add anthropics/claude-plugins-community
/plugin install trabuco@claude-community
You can also browse the catalog at claude.com/plugins, or pin to this repo directly (e.g. for unreleased changes) with /plugin marketplace add arianlopezc/Trabuco then /plugin install trabuco@trabuco-marketplace.
The plugin ships 8 skills (/trabuco:new-project, /trabuco:design-system, /trabuco:add-module, /trabuco:extend, /trabuco:migrate, /trabuco:doctor, /trabuco:suggest, /trabuco:sync), 17 specialist subagents (architect, AI-agent expert, migration orchestrator, 14 migration phase specialists), and an MCP server with 25 tools. Full plugin docs: plugin/README.md.
The plugin requires the trabuco CLI on PATH — install it first (or after; the SessionStart hook detects missing binaries and tells the assistant exactly what to do).
Highlights
- Multi-module Maven — clean compile-time boundaries between Model, SQLDatastore/NoSQLDatastore, Shared, API, Worker, EventConsumer, AIAgent.
- Spring Boot + Java — Spring Data JDBC (no JPA), Flyway migrations, virtual threads on by default, Testcontainers for real integration tests.
- OIDC Resource Server scaffolding (auto-generated for API/AIAgent) — Spring Security dual
SecurityFilterChain, JWT validation, scope-mapped authorities, RFC 7807 ProblemDetail handlers, RSA-signed e2e tests. Ships dormant — fliptrabuco.auth.enabled=trueand setOIDC_ISSUER_URIto validate tokens from Keycloak / Auth0 / Okta / Cognito / generic OIDC. No CLI flag, no half-installed projects. Full guide:docs/auth.md. - Production observability — RFC 7807 Problem Details, OpenTelemetry auto-instrumentation, Prometheus metrics, correlation IDs, health probes.
- AI Agent module — Spring AI with tools, LLM guardrails, MCP server, A2A protocol, multi-agent orchestration, knowledge base, webhooks. The OIDC chain coexists with the legacy
ApiKeyAuthFilter(governed by an independent property) for incremental migration. - AI collaboration layer —
.ai/prompts/task guides,JAVA_CODE_QUALITY.mdspec, per-agent rule files for Claude Code / Codex / Cursor / GitHub Copilot. - Migration of existing repos —
trabuco migrateruns a 14-phase orchestrated flow with per-phase approval gates and atomic rollback. - CI that knows your modules —
--ci githubemits a workflow with the exact Docker services your modules need. - Stay in sync with new releases —
trabuco sync --applyadds AI-tooling files added in newer CLI versions, additively.
For the full feature list, see docs/manual.md#features.
Documentation
docs/manual.md— full CLI reference (installation, modules, MCP server, code quality, configuration options, every flag).docs/migration-guide.md— migrating an existing Spring Boot project.docs/README.md— full docs index, including design context (philosophy, pattern recipes, when not to use, limitations).plugin/README.md— Claude Code plugin: skills, subagents, hooks, MCP server.
Requirements
- Java (Trabuco auto-detects installed versions)
- Maven
- Docker (for Testcontainers and local development)
Contributing
Issues and pull requests welcome.
License
MIT — do whatever you want with it.
Built with Trabuco. Now go build something amazing.