Azure on localhost:4577

A free, MIT-licensed Azure emulator with Azurite-compatible connection strings. Give your coding agents a real Azure to build and test against: no account, no subscription, no bill.

Get started floci-az on GitHub

# Install floci-cli, then start floci-az
curl -fsSL https://floci.io/install.sh | sh
floci az start
eval $(floci az env)

29 Azure services ready on :4577. Blob · Queue · Table · Functions · Key Vault · +24 more

24 ms startup · ~13 MiB idle · MIT licensed · no Azure account

Homebrew, Windows and Docker options

Quick Start

Every way to run it.

One CLI installs and runs all four emulators: use Homebrew, the install script or PowerShell. Or skip the CLI and run the Docker image.

1 · Install floci CLI

brew install floci-io/floci/floci

curl -fsSL https://floci.io/install.sh | sh

iwr https://floci.io/install.ps1 | iex

2 · Start the emulator

floci az start

3 · Export env vars

eval $(floci az env)
# exports AZURE_STORAGE_CONNECTION_STRING

4 · Upload a file

# Create a container
az storage container create \
  --name my-container \
  --connection-string $AZURE_STORAGE_CONNECTION_STRING

# Write a file and upload it
echo "Azure locally: no bill, no drama." \
  > hello-floci.txt
az storage blob upload \
  --container-name my-container \
  --name hello-floci.txt \
  --file hello-floci.txt \
  --connection-string $AZURE_STORAGE_CONNECTION_STRING

# Download it back and read it
az storage blob download \
  --container-name my-container \
  --name hello-floci.txt \
  --file hello-back.txt \
  --connection-string $AZURE_STORAGE_CONNECTION_STRING
cat hello-back.txt

Teach your coding agent

# Asks which agents to install it for
npx skills add floci-io/floci-cli

The Floci skill teaches Claude Code, Codex, Cursor, Kiro and other coding agents to run the emulators with floci-cli.

Other useful commands

# Check status
floci az status

# View logs
floci az logs

# Stop the emulator
floci az stop

# Print individual SDK endpoint vars
eval $(floci az env --format sdk-vars)
# exports AZURE_STORAGE_BLOB_ENDPOINT,
# AZURE_APP_CONFIGURATION_ENDPOINT,
# AZURE_KEY_VAULT_ENDPOINT, ...

Start with persistence

floci az start --persist ./data
# state survives container restarts

Diagnose issues

floci az doctor
# checks Docker, Azure CLI, connectivity

Examples

One blob upload, four SDKs.

Each sample reads AZURE_STORAGE_CONNECTION_STRING, which eval $(floci az env) sets. On the Docker path, export it from the Quick Start.

 # pip install azure-storage-blob
 import os

from azure.storage.blob import BlobServiceClient

client = BlobServiceClient.from_connection_string(

    os.environ["AZURE_STORAGE_CONNECTION_STRING"])

container = client.get_container_client("my-container")

if not container.exists():

    container.create_container()

blob = container.get_blob_client("hello.txt")

blob.upload_blob(b"Hello from floci-az!", overwrite=True)

 // azure-storage-blob:12.x
 BlobServiceClient client = new BlobServiceClientBuilder()

  .connectionString(System.getenv("AZURE_STORAGE_CONNECTION_STRING"))

  .buildClient();

BlobContainerClient container = client.getBlobContainerClient("my-container");

container.createIfNotExists();

BlobClient blob = container.getBlobClient("hello.txt");

blob.upload(BinaryData.fromString("Hello from floci-az!"), true);

 // Azure.Storage.Blobs NuGet package
 var client = new BlobServiceClient(

  Environment.GetEnvironmentVariable("AZURE_STORAGE_CONNECTION_STRING"));

var container = client.GetBlobContainerClient("my-container");

await container.CreateIfNotExistsAsync();

var blob = container.GetBlobClient("hello.txt");

await blob.UploadAsync(BinaryData.FromString("Hello from floci-az!"), overwrite: true);

 // npm install @azure/storage-blob
 import { BlobServiceClient } from "@azure/storage-blob";

const client = BlobServiceClient.fromConnectionString(

  process.env.AZURE_STORAGE_CONNECTION_STRING);

const container = client.getContainerClient("my-container");

await container.createIfNotExists();

const blob = container.getBlockBlobClient("hello.txt");

await blob.upload("Hello from floci-az!", 20);

floci-az · Azure services

29 Azure services. All of them free.

Wire-compatible with the official Azure SDKs and the Azurite emulator. Switch by changing the endpoint; your application code stays as it is.

Blob Storage REST XML / JSON Table Storage OData / REST JSON Cosmos DB SQL native · Mongo / Cassandra / Gremlin Azure SQL ARM · SQL Server Cache for Redis ARM · Redis PostgreSQL Flexible Server ARM · PostgreSQL MySQL Flexible Server ARM · MySQL MariaDB Server ARM · MariaDB Azure Functions HTTP / Timer Triggers Virtual Machines ARM · REST JSON AKS REST JSON · k3s Container Registry ARM · registry:2 Container Instances ARM · REST JSON Container Apps ARM · ingress proxy API Management ARM · REST JSON Virtual Network ARM · REST JSON Queue Storage REST JSON Event Hubs AMQP / Kafka / REST Service Bus AMQP · Topics / Queues Email Communication REST JSON · ARM Event Grid REST JSON · CloudEvents SignalR WebSocket · negotiate Microsoft Entra ID OAuth2 · OIDC · JWKS Managed Identity IMDS · OAuth2 tokens Microsoft Graph REST JSON · /v1.0 Key Vault REST · Secrets / Keys / Certs Azure Policy ARM · REST JSON Azure Monitor Logs · Metrics · KQL App Configuration REST JSON · Labels

Runs a Docker container Docker optional: mocked by default

Why floci-az

Azurite's connection string, and the rest of Azure.

Blob, Queue and Table answer the way Azurite does, next to Event Hubs, Service Bus, Cosmos DB, Key Vault and more on port 4577.

More than Azurite

Azurite covers Blob, Queue and Table. floci-az serves those with the same account name and key, plus 26 more services.

AMQP for native clients

Event Hubs and Service Bus also listen on AMQP ports 5672 and 5673, so native SDK clients connect without a proxy layer.

Real database engines

Azure SQL and the PostgreSQL, MySQL and MariaDB flexible servers run the real engines in Docker containers.

Native-image speed

Starts in 24 ms and idles at ~13 MiB, fast enough to start a fresh emulator for every test run.

From Azurite

Switching from Azurite?

Keep the account name, the key and the connection string format. Point the endpoints at port 4577 instead of Azurite's 10000, 10001 and 10002, and Queue and Table move to their own paths on the same port.

DefaultEndpointsProtocol=http;AccountName=devstoreaccount1;AccountKey=Eby8vdM02xNOcqFlqUwJPLlmEtlCDXJ1OUzFT50uSRZ6IFsuFq2UVErCz4I6tq/K1SZFPTOtr/KBHBeksoGMGw==;BlobEndpoint=http://127.0.0.1:10000/devstoreaccount1;QueueEndpoint=http://127.0.0.1:10001/devstoreaccount1;TableEndpoint=http://127.0.0.1:10002/devstoreaccount1;

DefaultEndpointsProtocol=http;AccountName=devstoreaccount1;AccountKey=Eby8vdM02xNOcqFlqUwJPLlmEtlCDXJ1OUzFT50uSRZ6IFsuFq2UVErCz4I6tq/K1SZFPTOtr/KBHBeksoGMGw==;BlobEndpoint=http://localhost:4577/devstoreaccount1;QueueEndpoint=http://localhost:4577/devstoreaccount1-queue;TableEndpoint=http://localhost:4577/devstoreaccount1-table;

Default credentials

The following fixed credentials match Azurite's defaults. All Azure SDK tooling expects these values for local development.

Property	Value
Account Name	`devstoreaccount1`
Account Key	`Eby8vdM02xNOcqFlqUwJPLlmEtlCDXJ1OUzFT50uSRZ6IFsuFq2UVErCz4I6tq/K1SZFPTOtr/KBHBeksoGMGw==`
Blob endpoint	`http://localhost:4577/devstoreaccount1`
Queue endpoint	`http://localhost:4577/devstoreaccount1-queue`
Table endpoint	`http://localhost:4577/devstoreaccount1-table`
Event Hubs AMQP	`amqp://localhost:5672`
Service Bus AMQP	`amqp://localhost:5673`
Event Hubs Kafka (opt-in)	`localhost:9093`

Azure, without the subscription.

MIT licensed and built for developer workflows. No Azure account, no subscription, no credit card. Just floci-cli and localhost.

floci-az on GitHub

Other emulators: AWS GCP OCI

license: MIT
real credentials: none
cloud account: none
cost: $0