Examples

Every example in examples/ is a deployable agent with a real agent.yaml and runnable code. Clone the repo and run any example locally in under 5 minutes.

git clone https://github.com/agentbreeder/agentbreeder
cd agentbreeder/examples/<example-name>
agentbreeder deploy --target local

Framework Examples

Start here if you want to see how a specific framework integrates with AgentBreeder.

LangGraph

examples/langgraph-agent/

Stateful multi-step research agent using LangGraph's graph-based orchestration. Demonstrates tool use, state management, and conditional edges.

cd examples/langgraph-agent && agentbreeder deploy --target local

CrewAI

examples/crewai-agent/

Role-based multi-agent crew: a Researcher agent and a Writer agent collaborating in sequence. Shows how CrewAI's crew definition maps to agent.yaml.

cd examples/crewai-agent && agentbreeder deploy --target local

Claude SDK

examples/claude-sdk-agent/

Claude SDK agent with adaptive thinking, prompt caching, and MCP tool use. Shows the claude_sdk: config block in agent.yaml.

cd examples/claude-sdk-agent && agentbreeder deploy --target local

OpenAI Agents SDK

examples/openai-agents-agent/

OpenAI Agents SDK with handoffs, guardrails, and file search. Shows how the Responses API maps to the AgentBreeder deploy pipeline.

cd examples/openai-agents-agent && agentbreeder deploy --target local

Google ADK

examples/google-adk-agent/

Google ADK agent with session memory and BigQuery grounding, deployed to GCP Cloud Run via AgentBreeder.

cd examples/google-adk-agent && agentbreeder deploy --target local

Cloud Target Examples

Use these to see how the same agent deploys to a specific cloud.

AWS App Runner

examples/aws-app-runner-agent/

Deploys to AWS App Runner — serverless containers with no VPC, no ALB, and no load balancer configuration required. Ideal for simple HTTP-accessible agents.

# agent.yaml
deploy:
  cloud: aws
  runtime: app-runner
  region: us-east-1

cd examples/aws-app-runner-agent
agentbreeder deploy --target app-runner --region us-east-1

Claude Managed Agents (Anthropic)

examples/claude-managed-agent/

Deploys to Anthropic's managed cloud — no container build, no infrastructure to provision. Anthropic manages the execution environment. Requires deploy.cloud: claude-managed and Claude 4.5+ model.

# agent.yaml
deploy:
  cloud: claude-managed

claude_managed:
  environment:
    networking: unrestricted
  tools:
    - type: agent_toolset_20260401

cd examples/claude-managed-agent
agentbreeder deploy --target claude-managed

Local Model Examples

Run agents with no API keys required.

Ollama (Local LLMs)

examples/ollama-agent/

Runs a full agent locally using Ollama and any GGUF-compatible model (Gemma, Llama, Mistral, Phi). No API key needed.

# Pull a model first
ollama pull gemma4

cd examples/ollama-agent
agentbreeder deploy --target local

# agent.yaml
model:
  primary: ollama/gemma4
  gateway: ollama

deploy:
  cloud: local
  runtime: docker

OpenRouter (300+ models, one API key)

examples/openrouter-agent/

Routes to DeepSeek-R1, Llama 3, Mistral, or 300+ other models via OpenRouter. One OPENROUTER_API_KEY instead of per-provider keys.

# agent.yaml
model:
  primary: openrouter/deepseek/deepseek-r1
  gateway: openrouter

agentbreeder provider add openrouter --api-key <your-key>
cd examples/openrouter-agent && agentbreeder deploy --target local

Multi-Agent Examples

A2A Subagent (Agent-to-Agent)

examples/a2a-subagent/

Coordinator + specialist subagent pattern using the A2A protocol. A research-coordinator agent dynamically delegates tasks to a web-search subagent and a summarizer subagent at runtime.

Shows how to:

• Configure a2a: in agent.yaml
• Register subagents in the org registry
• Route tasks between agents over JSON-RPC

cd examples/a2a-subagent
agentbreeder deploy --target local   # deploys all agents

Orchestration Patterns

examples/orchestration/

Four production-ready orchestration patterns as orchestration.yaml files:

cd examples/orchestration/fan-out-fan-in
agentbreeder orchestration deploy orchestration.yaml
agentbreeder orchestration chat research-pipeline

RAG and Knowledge Base Examples

GraphRAG with Ollama

examples/graphrag-ollama-agent/

Knowledge base + Neo4j graph RAG hybrid search, powered by local Ollama. No cloud API keys required. Includes an ingest.py script to populate the knowledge graph.

cd examples/graphrag-ollama-agent
python ingest.py                             # populate Neo4j knowledge graph
agentbreeder deploy --target local

Real-World Agent Examples

AI News Digest

examples/ai-news-digest/

Production-quality agent that runs on a daily cron schedule:

1. Fetches top stories from Hacker News, ArXiv (cs.AI + cs.LG), and RSS feeds
2. Synthesises a grouped digest using Google ADK + Ollama Gemma3:27b
3. Emails the digest to configured recipients via Gmail

Demonstrates: agentbreeder schedule, MCP tool server, SMTP connector, Google ADK framework, Ollama local model.

cd examples/ai-news-digest

# One-time setup
agentbreeder setup

# Run now
agentbreeder chat --agent ai-news-digest

# Schedule daily at 7am
agentbreeder schedule create --cron "0 7 * * *" --agent ai-news-digest

TypeScript SDK Examples

Basic TypeScript SDK

examples/sdk-ts-basic/

Minimal TypeScript SDK example — define an agent with @agentbreeder/sdk, add tools and guardrails, print the generated agent.yaml.

import { Agent, Tool } from "@agentbreeder/sdk";

const agent = new Agent("research-assistant", {
  version: "1.0.0",
  team: "engineering",
  framework: "claude_sdk",
})
  .withModel("claude-sonnet-4-6", { temperature: 0.7 })
  .withTool(searchTool)
  .withGuardrail("pii_detection")
  .withDeploy("aws", { runtime: "ecs-fargate" });

console.log(agent.toYaml());

cd examples/sdk-ts-basic
npm install && npm start

Advanced TypeScript SDK

examples/sdk-ts-advanced/

Multi-agent orchestration, memory configuration, MCP server wiring, and cloud deploy — all from TypeScript code.

cd examples/sdk-ts-advanced
npm install && npm start

Python SDK Examples

Basic Python SDK

examples/sdk-basic/

Minimal Python SDK example — define an agent programmatically, deploy to local Docker.

Advanced Python SDK

examples/sdk-advanced/

Full SDK usage: memory, knowledge bases, prompt registry references, MCP servers, and multi-cloud deploy targets — all from Python code.

Template Examples

examples/templates/

12 production-quality agent templates covering common enterprise use cases. Each is a complete agent.yaml with recommended model, tools, guardrails, and deploy config.

agentbreeder template list
agentbreeder template use customer-support --name my-support-agent
agentbreeder deploy --target local

All Examples at a Glance