Pick interfaces and backends, and get a complete Spring Boot or Quarkus project (JVM + native) with Apache Camel as the integration engine: routes, unit/integration/black-box tests, Docker Compose, and an embedded debugger UI — all wired and passing. You focus on business logic.
Initializer — a visual web app (like Spring Initializr, but for full Apache Camel projects with routes, tests, and Docker pre-wired). Pick options, click Generate, download a ready-to-run Maven project.
MCP Server — an endpoint your AI tool calls over the Model Context Protocol. Describe what you want, and the project appears in your workspace — no UI clicks.
A short walkthrough: generate a microservice, then watch messages flow through your Camel routes in real time with the embedded debugger — review what AI-generated code actually does, instead of reading it line by line.
Two phases: Generate your starting point, then Evolve it safely with AI as your needs grow.
Two ways to generate: the visual Initializer in your browser, or directly from your AI tool via MCP.
Pick interfaces (REST, Kafka, gRPC, MCP, etc.), backends, and CRUD operations — in the Initializer UI, or by describing what you need to your AI tool (which calls the CamelBee MCP Server). Choose Spring Boot or Quarkus (JVM + native). You get a complete Maven project: Camel routes, MapStruct mappers, 3-level tests, Docker Compose, k6 load tests, and AI-tool context files.
Launch Initializer
Open the project in Claude Code, Cursor, Codex, OpenCode, or Copilot and just ask: "Replace the Order domain with Payment", "Add credit-limit validation before the payment backend", "Run the tests". The AI follows the 3-layer architecture, Camel DSL patterns, and test patterns shipped with the project — routes, mappers, models, and unit + integration + black-box tests, all consistent with the existing code.
The embedded React UI runs directly inside your microservice. Watch messages flow through your Camel routes in real-time, inspect payloads, and trace transformations — no external tooling needed. Run /cb-run-tests to verify everything passes, then /cb-debug to launch the debugger.
You start from a working microservice — not a skeleton.
See it in action: Product Catalog MCP server — generated from the Initializer, then built by Claude Code in 21 minutes from a single prompt (zero manual corrections). Read the walkthrough →
Need Kafka, gRPC, or a new database months later? Type /cb-add-interface or /cb-add-backend in Claude Code (or invoke the equivalent MCP tool from Cursor, Codex, OpenCode, or Copilot). It calls the CamelBee MCP Server, generates a reference project, and surgically adds only the new pieces into your existing microservice.
Your AI tool follows the same architecture, same patterns, same test structure. New routes, models, mappers, Docker services, TestContainers configs, WireMock stubs, test data producers, and backend verifiers are all included in the reference — the AI tool copies them in, not invents them.
The embedded Metrics and Health UI run inside your microservice in any environment — no separate Prometheus/Grafana stack to set up. Track route throughput, error rates, and per-endpoint latencies, and check liveness/readiness from the same React UI you used to debug.
Claude Code, Cursor, Codex, OpenCode, or Copilot — any AI tool that supports the Model Context Protocol can call the CamelBee MCP Server directly. Just ask: "Generate a Spring Boot or Quarkus microservice with Apache Camel routing — REST interface, Kafka backend, called orders-api" — you get a complete Maven project (routes, tests, Docker, embedded debugger UI) delivered straight into your workspace.
MCP endpoint: https://www.camelbee.io/mcp (HTTP streamable)
Get your API key:
cb_… key that appears.Drop this into your .mcp.json (or equivalent config) — works for Cursor, Codex, OpenCode, Copilot:
{
"mcpServers": {
"camelbee": {
"type": "http",
"url": "https://www.camelbee.io/mcp",
"headers": {
"Authorization": "Bearer YOUR_CAMELBEE_API_KEY"
}
}
}
}
Claude Code users — use the mcp-remote bridge instead:
Claude Code's native HTTP MCP transport doesn't yet handle long-running tool calls; the mcp-remote npx bridge wraps the same endpoint over stdio and works today with no extra setup.
{
"mcpServers": {
"camelbee": {
"command": "npx",
"args": ["mcp-remote", "https://www.camelbee.io/mcp",
"--header", "Authorization: Bearer YOUR_CAMELBEE_API_KEY"]
}
}
}
Once you've opened a generated project, your AI tool can also call the MCP Server to add new interfaces/backends later without re-generating from scratch.
Rules, slash commands, and architecture context tuned to CamelBee — so once the project is open in your AI tool, it doesn't just read your code, it knows how to build on it. Pick which tools via the aiTools setting, or take them all.
| AI Tool | Files shipped |
|---|---|
| Claude Code | .claude/ · CLAUDE.md · .mcp.json |
| Cursor | .cursor/ · AGENTS.md |
| Codex | AGENTS.md |
| OpenCode | .opencode/ · AGENTS.md · .mcp.json |
| GitHub Copilot | .github/copilot-instructions.md |
Define interfaces, operations, and backends — CamelBee generates complete Camel routes with clean object mappings and extensible business logic stubs.
Debugger, Metrics, and Health UI run directly inside your microservice. Watch messages flow, view performance metrics, and debug live traffic — no external tools needed.
Choose your framework. Quarkus supports both JVM and native mode — native compiles to a native executable with sub-second startup, minimal memory, ready for serverless and Kubernetes. All tests are verified on native too.
All 19 interfaces and 19 backends ship with pre-built TestContainers configurations, WireMock stubs, Docker Compose services, test data producers, and backend verification utilities — verified on JVM and native executables. AI doesn't need to invent how to test Kafka, Cassandra, gRPC, or DynamoDB — it's all there, proven, and passing.
Choose your framework in the Initializer. Quarkus supports both JVM and native mode — same generated code, two deployment targets.
The most widely adopted Java framework. Mature ecosystem, broad library support, familiar to most teams. Full JVM with standard startup and memory profile.
Kubernetes-native Java framework. Faster startup than Spring Boot, lower memory usage, live reload with quarkus:dev. CDI-based dependency injection. Great for development and traditional deployments.
Sub-second startup. Minimal memory. The same Quarkus project compiles to a native executable via GraalVM — ideal for serverless (AWS Lambda, Azure Functions, Google Cloud Run) and Kubernetes scale-to-zero. All 19 interfaces and 19 backends are tested and verified on native.
CamelBee handles native compilation requirements automatically — reflection registration, serialization configs, and native-compatible Camel DSL patterns are all included in the generated code. No extra configuration needed to go native.
| Interfaces | Backends |
|---|---|
|
Currently Supported:
Coming Soon:
|
Currently Supported:
Coming Soon:
|
Example projects built from the CamelBee Initializer — full source on GitHub, with walkthrough posts on Medium.
📋 Prompt Recipes — Build Microservices with One Prompt
Copy-paste prompts for real use cases: Product Catalog MCP Server, Currency Exchange Rate, Loan Application Processing. Step 1: generate via CamelBee. Step 2: paste into your AI tool. No IDE required.
One prompt, zero manual corrections: Claude Code transformed a CamelBee-generated Quarkus project into a Product Catalog MCP server with audit logging, full test coverage, and native build — driven entirely by the structured guardrails in the generated CLAUDE.md.
A raw CamelBee-generated project, pushed to GitHub without a single modification, used to benchmark gRPC vs REST with Protocol Buffers vs REST with JSON. Out-of-the-box performance numbers from the generator — no optimization work required.
Why reading AI-generated code line by line doesn't scale — and how the embedded CamelBee Debugger lets you review what a microservice actually does by watching messages flow through its Camel routes in real time.
Read on Medium →Stay up to date with the latest CamelBee tutorials, integration patterns, and deep-dives into Apache Camel microservice development — published regularly on Medium.
Read on Medium →You don't need to start from scratch! Add CamelBee's core libraries to your existing Apache Camel microservices and instantly unlock the Debugger, Metrics, and Health UI — all as a React embedded UI running directly inside your microservice.
Add the Spring Boot core library to your existing Camel application and start visualizing your routes immediately.
View Spring Boot Core →Integrate the Quarkus core library into your Camel Quarkus services for real-time debugging and performance monitoring.
View Quarkus Core →Simply add the dependency and your existing microservices will have the full CamelBee React embedded UI — Debugger, Metrics, and Health UI — for live debugging and monitoring
Generate today, evolve tomorrow. Start with the Initializer — add new technologies anytime from your AI tool.