Examples Guide

Chronos ships with 25+ runnable examples covering every major feature. Most require no API keys and run entirely with mock providers and SQLite. They are grouped by type so you can jump straight to what you need:

Category	What it covers
Fundamentals	Agent builder, tools, guardrails, graph patterns, memory, streaming — all no-key
LLM Agents	Live-LLM workflows: StateGraph reasoning, MCP tools, coding agent, multi-agent teams
Providers & Models	Azure OpenAI, Vertex AI, Bedrock, multi-provider comparison, failover
Durability & Sandboxing	Durable queue, human-in-the-loop resume, process sandboxing
Enterprise & Multi-Tenancy	OIDC/JWKS SSO, data residency, per-tenant memory isolation
Observability & CLI	Metrics/cost/cache/retry hooks, CLI-driven agents and ops
YAML Agent Examples	Progressive YAML applications: single agents, routers, pipelines, advanced teams, and governed deployments

Running Examples

# Clone the repo
git clone https://github.com/spawn08/chronos.git
cd chronos

# Run any example
go run ./examples/<name>/

Choosing a provider

Every example that talks to a real LLM (graph_with_llm, mcp_agent, coding_agent, team_deploy, multi_agent, multi_provider) resolves its provider through a shared helper — examples/internal/providers.Pick() — which reads the environment and returns the first configured provider. You never edit the example to switch clouds; you just export a different set of environment variables.

Provider	Environment variables
OpenAI	OPENAI_API_KEY
Anthropic	ANTHROPIC_API_KEY
Google Gemini (AI Studio)	GEMINI_API_KEY
Azure OpenAI	AZURE_OPENAI_API_KEY + AZURE_OPENAI_ENDPOINT + AZURE_OPENAI_DEPLOYMENT (+ optional AZURE_OPENAI_API_VERSION)
Google Cloud Vertex AI	GOOGLE_CLOUD_PROJECT + GOOGLE_ACCESS_TOKEN (+ optional GOOGLE_CLOUD_LOCATION, VERTEX_MODEL)
AWS Bedrock	AWS_REGION + AWS_ACCESS_KEY_ID + AWS_SECRET_ACCESS_KEY (+ optional BEDROCK_MODEL_ID)
Mistral	MISTRAL_API_KEY
Ollama (local)	OLLAMA_HOST (+ optional OLLAMA_MODEL)
Any OpenAI-compatible	OPENAI_COMPATIBLE_BASE_URL + OPENAI_COMPATIBLE_MODEL (+ optional _API_KEY, _NAME) — Together, Groq, DeepSeek, OpenRouter, Fireworks, Perplexity, Anyscale, vLLM, LiteLLM

The four common cloud setups look like this — pick one, then run any real-LLM example:

# OpenAI
export OPENAI_API_KEY=sk-...

# Azure OpenAI
export AZURE_OPENAI_API_KEY=...
export AZURE_OPENAI_ENDPOINT=https://your-resource.openai.azure.com
export AZURE_OPENAI_DEPLOYMENT=gpt-4o          # your deployment name
export AZURE_OPENAI_API_VERSION=2024-12-01-preview

# Google Cloud Vertex AI (OpenAI-compatible endpoint + gcloud token)
export GOOGLE_CLOUD_PROJECT=my-gcp-project
export GOOGLE_CLOUD_LOCATION=us-central1
export VERTEX_MODEL=google/gemini-2.5-pro
export GOOGLE_ACCESS_TOKEN=$(gcloud auth print-access-token)

# AWS Bedrock
export AWS_REGION=us-east-1
export AWS_ACCESS_KEY_ID=...
export AWS_SECRET_ACCESS_KEY=...
export BEDROCK_MODEL_ID=anthropic.claude-3-5-sonnet-20241022-v2:0

tip

Precedence is fixed (OpenAI → Anthropic → Gemini → Azure → Vertex → Bedrock → Mistral → Ollama → compatible). To force a specific cloud when several key sets are present, unset the higher-precedence ones in that shell.

Full Index

Fundamentals — no API key

Example	Feature area
quickstart	Agent builder, SQLite, StateGraph
chat_with_tools	Tool calling loop
tools_and_guardrails	Permissions, guardrails
graph_patterns	Conditional edges, interrupts, streaming
memory_and_sessions	Short/long-term memory, sessions
streaming_sse	Event broker, SSE

LLM Agents — API key (mock fallback)

Example	Feature area
graph_with_llm	StateGraph + live LLM
mcp_agent	Model Context Protocol tools
coding_agent	Autonomous coding agent + RAG
multi_agent	4 team strategies, bus delegation
team_deploy	YAML teams + sandbox deploy

Providers & Models

Example	Feature area
multi_provider	Multiple providers side by side
azure	Azure OpenAI (chat + streaming)
vertex	Google Cloud Vertex AI
fallback_provider	Provider failover

Durability & Sandboxing — no API key

Example	Feature area
durable_queue	Leased workers, sleep, park/signal HITL, orphan recovery
durable_hitl	Human-in-the-loop approval with checkpoint + resume
sandbox_execution	Process sandbox with timeouts and I/O capture

Enterprise & Multi-Tenancy

Example	Feature area
enterprise_sso	ChronosOS behind OIDC/JWKS SSO
data_residency	Per-tenant storage routing (EU vs US)
multitenant_memory	Per-tenant long-term memory isolation on one agent

Observability & CLI

Example	Feature area
hooks_observability	Metrics, cost, cache, retry, rate limit
cli_agent	Build, inspect, run an agent from YAML via the CLI
cli_ops	Operate Chronos from the CLI (serve, monitor, db, sessions, pipe, deploy)