Claude Code Skills

Genkit Java ships with Claude Code skills — structured knowledge files that teach Claude Code how to work effectively with the framework.

Available skills

Skill	Description	Audience
building-ai-apps-with-genkit-java	Build AI-powered apps with Genkit Java: setup, flows, generation, tools, RAG, prompts, agents, deployment	App developers
developing-genkit-java	Contribute to the framework: architecture, plugin development, core abstractions, conventions	Framework contributors

Installation

Option 1: Install from GitHub

Install a specific skill:

# For app developers
npx skills add https://github.com/genkit-ai/genkit-java --skill building-ai-apps-with-genkit-java

# For framework contributors
npx skills add https://github.com/genkit-ai/genkit-java --skill developing-genkit-java

Skills are sourced from the skills/ directory and installed into your local Claude Code configuration.

Option 2: Manual setup

Copy the skill folders into your Claude Code skills directory:

# Clone the repo (if not already)
git clone https://github.com/genkit-ai/genkit-java.git

# Copy skills to your Claude Code config
cp -r genkit-java/skills/* ~/.claude/skills/

Usage

Once installed, Claude Code automatically activates the relevant skill based on your prompts. For example:

“Create a new Genkit Java app with OpenAI that summarizes text” → triggers the building-ai-apps-with-genkit-java skill
“Add a new Anthropic plugin to the Genkit Java framework” → triggers the developing-genkit-java skill

You can also invoke a skill explicitly in Claude Code:

/skill building-ai-apps-with-genkit-java "Build a RAG pipeline with Pinecone and Gemini"

What’s inside

building-ai-apps-with-genkit-java

Covers everything an app developer needs:

Quick start — minimal pom.xml, application code, run commands
Provider setup — all 12+ providers with Maven coordinates, env vars, model names
Generation API — text, streaming, structured output, multi-turn
Flows — defining, exposing as HTTP endpoints
Tools — function calling with typed I/O
DotPrompt — .prompt files with Handlebars templates and variants
RAG — embeddings, indexing, retrieval with local and production vector stores
Sessions & Chat — multi-turn conversations with state
Agents — multi-agent orchestration
Interrupts — human-in-the-loop patterns
Evaluations — built-in RAGAS metrics and custom evaluators
Deployment — Jetty, Spring Boot, Firebase Cloud Functions
Observability — tracing, metrics, Firebase telemetry

developing-genkit-java

Covers framework internals and contribution guidelines:

Module architecture — core, ai, genkit, plugins dependency flow
Core abstractions — Action<I,O,S>, ActionContext, Registry, ActionType
Plugin system — Plugin interface, registration, compat-oai base
Model implementation — Model interface, streaming, SSE
Naming conventions — packages, classes, methods, action keys
Testing patterns — JUnit 5, Mockito, test structure
Code quality — Google Java Format, Checkstyle
Conventional commits — types, scopes, versioning
New module checklist — step-by-step guide for adding plugins