Getting Started — Camel-Kit

Overview

Camel-Kit is an AI-guided integration development tool that transforms how you build Apache Camel integrations. Instead of writing code directly, you describe your integration requirements in natural language, and the AI handles design, planning, implementation, and verification.

Installation Walkthrough

Follow these steps to install Camel-Kit and create your first integration project.

Prerequisites

Before installing Camel-Kit, ensure you have:

Java 17 or higher - Required for running Apache Camel and JBang
JBang - CLI launcher for Camel-Kit
Camel JBang 4.18.0 or higher - Required for the camel kit plugin channel and complete route execution and startup checks
Camel JBang test plugin - Required for Citrus integration tests; install it with camel plugin add test
Docker (conditional) - Required for external-service environment probes and full Citrus/Testcontainers integration verification; design, planning, and non-container checks work without it

Verify Java:

java -version

Install JBang:

Visit jbang.dev or use your package manager:

# macOS
brew install jbangdev/tap/jbang

# Linux
curl -Ls https://sh.jbang.dev | bash -s - app setup

# Windows
choco install jbang

# Install Camel JBang and its test plugin
jbang app install camel@apache/camel
camel plugin add test

Install Camel-Kit

Current development channel: JBang App Install (Recommended)

The easiest way to install Camel-Kit:

jbang app install camel-kit@luigidemasi/camel-kit

This adds the camel-kit command to your PATH. The GitHub alias installs the latest deployed 0.3.2-SNAPSHOT, which may lag main until the next deployment. Build from source when you need the exact current revision; both current channels provide the commands and AI targets described by this site.

Verify installation:

camel-kit --version

Run without installing:

jbang run camel-kit@luigidemasi/camel-kit init ...

Stable channel: Camel JBang plugin 0.3.1

Maven Central currently provides the stable 0.3.1 plugin. Pin that version explicitly:

camel plugin add kit \
  --gav io.github.luigidemasi:camel-jbang-plugin-kit:0.3.1 \
  --description "Design Apache Camel Integrations with AI"

Stable 0.3.1 exposes only camel kit init and the bob, gemini, and claude targets. It does not provide the current 0.3.2-SNAPSHOT command or nine-agent surface; do not use a dynamic Maven version when you need current-source behavior.

Initialize Project

Initialize Your First Project

Create a new integration project:

camel-kit init my-project --ai claude
cd my-project

The init command checks for prerequisites (Java 17+, JBang, Camel JBang, Camel test plugin) and reports their status. If the directory already contains a camel-kit project, it warns and exits — use --force to overwrite.

What gets created:

AGENTS.md - Routing table for AI agents
Agent-native skills and entry points - for example, .claude/commands/ for Claude Code or .agents/skills/ for Codex
docs/constitution.md - Generated architecture rules
docs/flows/, test/data/, and schemas/ - Empty initialization scaffolds; generated pipeline artifacts use the selected runtime’s paths
.camel-kit/ - Project configuration, cached catalogs, templates, and pipeline state
Target-specific MCP configuration - JSON for most targets, .codex/config.toml for Codex, and .mcp.json consumed through pi-mcp-adapter for Pi
mvnw, mvnw.cmd, and .mvn/wrapper/maven-wrapper.properties - Maven wrapper launchers and configuration

Choose your AI agent:

Agent	Flag
IBM Bob 2	`--ai bob2` (default)
IBM Bob 1	`--ai bob` (legacy)
Gemini CLI	`--ai gemini`
Claude Code	`--ai claude`
OpenAI Codex CLI	`--ai codex`
GitHub Copilot CLI	`--ai copilot`
Pi	`--ai pi` (then `pi install npm:pi-mcp-adapter@2.11.0`)
Qwen Code	`--ai qwen`
OpenCode	`--ai opencode`

First Command

Run Your First Command

For greenfield development:

Open your AI coding agent in the project directory and say:

I want to build an integration that reads from Kafka and writes to a database

The AI reads AGENTS.md, uses /camel-start to route the request to /camel-brainstorm, and guides you through design.

On the first run, no active pipeline ID exists yet. The agent asks you to create one before the interview continues (see camel-kit nextId in the command reference):

camel-kit nextId kafka-to-database
# Camel JBang plugin equivalent: camel kit nextId kafka-to-database

With OpenAI Codex CLI, run /skills and invoke $camel-start instead; Codex discovers .agents/skills/ directly and does not use slash-command stubs. See the Codex setup guide.

For migration:

camel-kit init --here --ai claude --source-platform mulesoft

Then tell your AI agent:

Migrate my MuleSoft flows to Camel

The AI invokes /camel-migrate and handles the conversion.

What's Next

What’s Next?

Now that you have Camel-Kit installed and a project initialized:

Learn the pipeline:

Read the Pipeline Overview to understand the four stages

Choose your workflow:

Greenfield Workflow - Build integrations from scratch
Migration Workflow - Convert existing integrations

Configure your AI target:

OpenAI Codex CLI - Repository trust, skills, custom agents, MCP, approvals, and sandboxing

Explore features:

Graph CLI - Migration analysis tools
Knowledge - Q&A about Camel components

Get help:

Documentation at this site
GitHub: github.com/luigidemasi/camel-kit
Community discussions in the repository

Two Workflows

Camel-Kit supports two integration workflows. Toggle to see each approach:

Greenfield Development

Build new integrations from scratch using the Design → Plan → Execute → Validate pipeline:

Run /camel-brainstorm — AI interviews you about requirements
Review and approve the design specification
AI automatically decomposes into tasks and generates code
AI verifies the integration with runtime tests
AI runs the final static quality gate and reports findings without modifying routes

camel-kit init my-project --ai claude
# Then in your AI agent:
# "I want to build a Kafka-to-PostgreSQL integration"

→ Greenfield Workflow guide

Platform Migration

Migrate existing integrations from other platforms:

MuleSoft Mule 3.x/4.x — Analysis of Mule XML flows, DataWeave, and connectors
Microsoft BizTalk — Orchestration, map, and pipeline migration
Apache Camel 2.x/3.x — Modernize to Camel 4.x YAML DSL
JBoss Fuse — Legacy Fuse/Karaf migration; Fuse projects are Camel 2.x/3.x codebases, so use --source-platform camel (or let auto detect the Fuse version qualifiers)

The AI auto-detects the source platform, records flow-specific analysis in one migration design package, and—after one approval—plans and executes the complete migration in dependency waves. It then runs one project-wide runtime verification pass and the final report-only validation gate.

camel-kit init --here --ai claude --source-platform mulesoft
# Then in your AI agent:
# "Migrate my MuleSoft flows to Camel"

→ Migration Workflow guide