Floci: The Free, Open-Source Local AWS Emulator That Replaces LocalStack
Floci is a free, open-source local AWS emulator built with Quarkus Native. A drop-in replacement for LocalStack that requires zero configuration, zero authentication, and runs 75+ services on a single port.
Why Local Developers Need a Local Cloud
Every cloud-native developer has been here: you write code that talks to S3, DynamoDB, SQS, or Lambda. It works in production. Then you push to your local machine and everything breaks — because those services don't exist locally.
For years, the answer was LocalStack. It was the de facto standard for running AWS services on your laptop. But in 2024, LocalStack made a controversial decision: they removed their free community edition, leaving developers who couldn't or wouldn't pay without a local cloud option.
Enter Floci — a free, open-source, MIT-licensed local cloud emulator that was created specifically to fill this gap. Named after cirrocumulus floccus (a small, fluffy cloud formation), Floci is a drop-in replacement for LocalStack that requires zero configuration, zero authentication tokens, and zero compromises.

Cloud computing infrastructure. Image credit: Open Grid Scheduler / Grid Engine, Wikimedia Commons (CC0).
What Is Floci?
Floci is a local AWS emulator written in Java using Quarkus and compiled to a GraalVM Mandrel native binary. It runs as a single process on port 4566 — the same port LocalStack uses — so switching requires zero changes to your existing code or tooling.
As of mid-2026, Floci supports 75+ AWS services across compute, storage, databases, messaging, identity, infrastructure, and container-backed workloads. It also extends beyond AWS: Azure, GCP, and OCI emulation is available through companion modules.
Key Technical Details
| Property | Value |
|---|---|
| License | MIT |
| Language | Java (Quarkus) |
| Build | GraalVM Mandrel native binary |
| Default Port | 4566 (AWS-compatible wire protocol) |
| Services | 75+ AWS, plus Azure/GCP/OCI |
| Auth Required | None — no account, no API keys |
| Docker Support | Yes — single docker compose up |
How It Works
Floci's architecture is deceptively elegant. Behind the scenes, it implements the real AWS wire protocol using a Smithy-based service descriptor system. Here's what that means in practice:
Smithy-Spec Protocol Routing
Floci doesn't fake responses with hardcoded stubs. It uses Smithy service specifications to define the actual API surface of each emulated service. When your code sends an s3:GetObject request, Floci parses it against the real S3 Smithy spec, routes it through the correct handler, and returns a response that matches what the real AWS SDK expects.
This is fundamentally different from older approaches that relied on regex-based URL matching or manual mock implementations. The Smithy spec approach means:
- Protocol fidelity: Content-type headers, request IDs, and error codes match real AWS
- SDK compatibility: Any AWS SDK (v2, v3, Boto3, Terraform, CDK) works without modification
- Future-proofing: New AWS features are supported by updating the Smithy spec, not rewriting handlers
Four Storage Modes
Floci supports multiple backends for emulated service data:
- In-memory — fastest option, perfect for unit tests and CI pipelines where persistence doesn't matter
- File-based — stores data on disk as JSON files, useful for debugging and reproducible test scenarios
- SQLite — persistent relational storage, good for medium-sized test datasets
- PostgreSQL — production-grade persistence, allows you to inspect emulated data with SQL queries
This flexibility means you can run Floci at full speed during local development and switch to PostgreSQL-backed mode when you need to share test fixtures across a team.
Multi-Cloud Support
While AWS is the primary focus, Floci has expanded beyond a single-cloud emulator:
- Azure: Emulates core services like Blob Storage, Service Bus, Key Vault, and Functions
- GCP: Supports Cloud Storage, Pub/Sub, and BigQuery local emulation
- OCI (Oracle Cloud): Provides Object Storage and Functions emulation
This makes Floci a genuine multi-cloud development platform rather than just an AWS clone.
Getting Started
Docker Compose (Recommended)
The fastest way to run Floci is with Docker Compose. Create a docker-compose.yml:
services:
floci:
image: flociio/floci:latest
ports:
- "4566:4566"
environment:
- FLOCI_SERVICES=s3,lambda,sqs,dynamodb,iam,cloudformation
Then run:
docker compose up -d
That's it. No API keys. No AWS account. No configuration.
AWS CLI Configuration
Floci uses placeholder credentials that work with any AWS SDK or CLI tool:
export AWS_ACCESS_KEY_ID=fake
export AWS_SECRET_ACCESS_KEY=fake
export AWS_DEFAULT_REGION=us-east-1
export AWS_ENDPOINT_URL=http://localhost:4566
Verify It Works
aws --endpoint-url http://localhost:4566 s3 ls
# Returns empty list — S3 is ready and working
Floci vs LocalStack
The comparison table below shows why many developers have migrated:
| Feature | Floci | LocalStack (Free Tier) |
|---|---|---|
| License | MIT | Proprietary (free tier removed) |
| Cost | Free, always | Paid subscription required |
| Services | 75+ AWS + Azure/GCP/OCI | ~50 AWS services |
| Startup Time | ~2 seconds (native binary) | ~10-30 seconds (JVM) |
| Memory Footprint | ~150MB | ~500MB+ |
| Authentication | None required | Required for paid tier |
| Persistence | In-memory, file, SQLite, PostgreSQL | Limited |
| Community Edition | Full open source | Removed in 2024 |
Why Floci Is Faster
Floci's speed advantage comes from its GraalVM Mandrel native binary. Unlike LocalStack which runs on the JVM (requiring a full Java runtime and garbage collection), Floci is compiled ahead-of-time to native machine code. This means:
- Near-instant startup — no JVM warmup
- Lower memory footprint — no garbage collector overhead
- Faster request processing — direct native code execution
In practice, this translates to faster CI pipelines (tests spin up in seconds instead of minutes) and a lighter local development environment.
Real-World Use Cases
Local Development
Developers can build and test full cloud-native applications locally without hitting AWS rate limits or incurring costs. Spin up S3, DynamoDB, SQS, Lambda, and API Gateway — all on localhost — and iterate as fast as your code changes.
CI/CD Pipelines
Floci's in-memory mode makes it ideal for CI environments. Start the emulator, run your integration tests, tear it down — all in under 30 seconds. No external service dependencies, no flaky network calls to real AWS.
AI Coding Agents
Floci explicitly positions itself as a tool for AI coding agents. Tools like Claude Code, Codex, and Cursor can safely interact with emulated services during development without risking production credentials or hitting rate limits. The zero-authentication model means agents don't need API keys to experiment.
Education and Prototyping
Students and architects can prototype cloud architectures without creating an AWS account. Spin up a complete event-driven system with SQS, Lambda, and DynamoDB in minutes, not hours.
A Complete Example: Event-Driven Architecture
Here's how you'd build a complete event-driven AWS architecture locally with Floci:
# docker-compose.yml
services:
floci:
image: flociio/floci:latest
ports:
- "4566:4566"
environment:
- FLOCI_SERVICES=sqs,s3,lambda,dynamodb
# Create an SQS queue
import boto3
sqs = boto3.client('sqs', endpoint_url='http://localhost:4566')
queue_url = sqs.create_queue(QueueName='orders')['QueueUrl']
# Publish a message
sqs.send_message(QueueUrl=queue_url, MessageBody='{"order_id": 123}')
# Read the message
messages = sqs.receive_message(QueueUrl=queue_url)
print(messages['Messages'][0]['Body'])
# → {"order_id": 123}
This works identically with the real AWS SDK — no code changes needed. When you're ready to deploy, just change the endpoint_url and everything else stays the same.
Limitations and Considerations
Service Coverage
While Floci covers the vast majority of commonly-used AWS services, some edge-case or newer services may not be fully implemented yet. Always verify your specific service needs against the official documentation.
Data Persistence
In-memory mode (the default) loses all data when the container restarts. For persistent local development, use PostgreSQL-backed mode. Be aware that emulated storage is not a replacement for production infrastructure — it's a development tool.
Performance
While Floci is significantly faster than LocalStack, emulated services will never match the performance of real AWS. Don't use Floci for load testing or performance benchmarking.
Multi-Cloud Modules
Azure, GCP, and OCI support is newer than the AWS implementation. Some services may have limited coverage compared to the core AWS emulation.
The Bigger Picture: Why Free Local Emulators Matter
Floci's existence addresses a fundamental tension in cloud-native development: developers need local service parity for productivity, but cloud providers want to lock them into their ecosystems. When LocalStack removed its free tier, it left a vacuum that Floci filled almost overnight.
The project's timing was no accident. It opened one week after LocalStack archived its community edition, and the response from the developer community was immediate. Within weeks, it had become one of the most talked-about tools in the Quarkus ecosystem — featured on Quarkus Insights #256.
Floci represents a broader trend: the democratization of cloud development tooling. When you can run 75+ cloud services locally with zero credentials and zero cost, the barrier to entry for cloud-native development drops dramatically.
Final Thoughts
If you're still paying for LocalStack or juggling real AWS accounts for local development, Floci is worth trying. The migration is trivial — same port, same API, same SDKs — and the performance improvement is noticeable from the moment you run docker compose up.
It's free, open-source, MIT-licensed, and actively maintained. In a world where local cloud tooling has become expensive and restricted, Floci is a breath of fresh air.
Resources:
Tags: AWS, local-development, open-source, cloud-native, developer-tools, floci, LocalStack, Quarkus, DevOps
Feature image: Open Grid Scheduler / Grid Engine / CC0 / Wikimedia Commons.