· via Hacker News – Front Page (hnrss.org)
Floci ships MIT-licensed local emulators for AWS, Azure, GCP and Oracle Cloud
Floci, which hit Hacker News's front page, distributes standalone MIT-licensed binaries that emulate AWS, Azure, GCP and Oracle Cloud services locally, with no accounts, tokens or telemetry.

What Floci is
Floci, a project that reached the front page of Hacker News, publishes four standalone binaries that emulate cloud platforms on a developer's own machine: floci for AWS, floci-az for Azure, floci-gcp for Google Cloud and floci-oci for Oracle Cloud. According to the project's homepage, each binary is MIT-licensed and needs no cloud account, API key or sign-up; services run entirely on localhost, with each platform's services exposed behind a single port.
Coverage varies by cloud. The AWS emulator spans 119 services, including S3, SQS, Lambda, DynamoDB, RDS and EKS. The Azure binary covers 29 services, among them Blob, Queue, Table, Functions, App Config, Key Vault, Event Hubs and Service Bus. The GCP binary offers 25 services such as GCS, Pub/Sub, Firestore, Cloud Run, Cloud SQL and GKE, while the OCI binary handles eight, including Object Storage, Identity, Queue, Streaming, KMS, Vault and Functions.
A LocalStack-shaped wedge
The AWS emulator is explicitly positioned as a swap for LocalStack: it listens on the same port 4566, so existing tooling can point at it without code changes. Floci's homepage asserts that LocalStack began requiring an auth token in March 2026 and commits to never doing the same; that assertion about LocalStack comes from Floci's own marketing and is not verified here. Integration relies on standard endpoint overrides such as AWS_ENDPOINT_URL pointing at localhost, which the site says works with every SDK, CLI, Terraform or OpenTofu module and test runner without special adapters.
Performance and implementation claims
Floci reports a 24 ms cold start and 13 MiB of idle memory, attributing this to native compilation with GraalVM Mandrel. The project also says it runs genuine engines rather than stubs: Lambda functions execute in real Docker containers, the RDS emulation is backed by actual PostgreSQL and MySQL, and ElastiCache is real Redis. The site claims full protocol fidelity, so code an agent verifies locally should behave the same way in production. All of these figures come from the project itself; no independent benchmarks accompanied the launch.
Built for AI agents
A large part of the pitch targets AI coding agents. Agents can produce cloud code quickly, but running it against live accounts risks leaking credentials and racking up charges. Floci's answer is to have agents connect with disposable keys, keep real secrets out of the agent's context and sandbox, and confine the worst-case failure to resetting a local container instead of damaging a staging environment. The fast startup is presented as letting agents build, test and tear down environments within a normal coding iteration rather than waiting for a remote account to respond.
Where it fits
The site lists four main use cases: ephemeral per-job test environments inside CI pipelines, offline local-first development against services like S3 or Service Bus, dry-running Terraform, OpenTofu or CloudFormation changes against localhost before touching real infrastructure, and running workshops where every participant gets a full stack with no billing risk. A unified CLI and a visual dashboard manage all four emulators. Installation is a single curl script, after which the published quick start demonstrates creating an S3 bucket and copying a file through it using the standard AWS CLI.
Why it matters
Local emulation has quietly become core infrastructure for cloud teams, and licensing shifts around established tools create space for a fully open alternative. Floci's combination of MIT licensing, no accounts and no telemetry, plus native-binary startup speeds, lowers the barrier for both laptops and CI — and its approach of removing credentials from the loop maps neatly onto how AI agents are increasingly asked to verify the code they write. The caveats are real, though: service counts, performance numbers and fidelity claims all come from the project's own page, and parity across 119 AWS services will need validation from early users. If the emulation holds up in practice, a free, forkable local cloud is a useful addition to the development toolkit.
- #cloud-emulation
- #local-development
- #open-source
- #aws
- #devtools