A cross-platform, forkable, time-traveling distributed filesystem over object storage.

Mount it natively on Mac, Windows, Linux, and Kubernetes, or reach it over S3 and WebHDFS, on owned infrastructure.

Get started(opens in new tab)
Time travelOpen the volume exactly as it existed at any point in time.Tue 14:32Wed 09:101 hr ago10 min agonowLocal file accessWork with familiar files and folders.▾ project-alphadatasetsnotesspec.mdbuildForksFork a volume instantly. Work inisolation, keep it or drop it.mainforkVersion historyEvery file keeps a rolling historywith one-minute granularity.some.filev1v2v3v4v5Recently deletedCopy back files removed by accident,within the retention window.old-report.csvDeleted 20 min ago
Why mountOS

Built in, not bolted on.

Local file access, time travel, forks, version history, and immutable data are native to the storage itself.

Works anywhere

Runs everywhere

Mount it natively on Mac, Windows, and Linux, or reach the same data over the S3 API and HDFS. One volume, every surface.

Standard semantics

POSIX-compliant

A fully POSIX-compliant filesystem. Standard file and directory semantics let existing applications and tools run as they are written.

No copy

Forks

Fork an entire volume in one call. Work in isolation, fork again, then keep it or drop it. Nothing is copied.

Ephemeral

Sandbox mounts

Mount a temporary, writable fork of a volume, pinned to any point in its history or its live state. Experiment freely, then unmount and every change is discarded.

Point in time

Time travel

Open the volume as it was a minute ago, an hour ago, or last week. Any minute inside the retention window.

Sub-minute granularity

Version history

Every file keeps a rolling history automatically, one minute by default, configurable down to a sub-minute window. Rapid saves collapse to the latest one so the history stays useful.

Built-in recovery

Recently deleted

Open files removed by accident and copy them back out, from the volume itself.

Per volume

Retention policy

One setting per volume controls how far back time travel reaches, how long deleted files stay available, and when old versions age out.

SDK

Integration

An open-source dashboard and SDK for custom integration, with scoped access tokens.

Model Context Protocol

AI MCP connector

Each client ships an MCP connector so AI agents can work with mounted volumes.

Isolated plane

Metadata

Metadata runs on its own isolated plane, served from an in-memory layer and persisted to PostgreSQL or MySQL-wire-compatible engines.

Protocols

Protocol support

Mount FUSE · NFS · SMB · FSKit · mountosio · Kubernetes CSI

Compatible POSIX · S3 · WebHDFS

Plans

Self-host free. Talk to us to scale.

Every plan carries every capability. Reach out to grow past the free tier or to add support.

Free

Self-hosted

Free / up to 10 TiB

The entire suite runs on owned infrastructure.

  • Every capability
  • All client surfaces (mounts, S3, WebHDFS, CSI)
  • Forks, time travel & version history
  • Single HUB deployment
Get started(opens in new tab)
Scale

Business License

Let's talk

Room to grow, arranged with us directly.

  • Storage beyond the free tier
  • Priority support and licensing
Contact us

support@mountos.io

FAQ

Questions, answered.

Why mountOS?

mountOS mounts object storage as a native filesystem. It lets infrastructure providers deliver production-ready native file access as a product of their own.

Where does my data live?

On owned infrastructure: owned servers, and owned object storage.

How is it licensed?

Free up to 10 TiB for a single self-hosted deployment, with every capability included. Beyond that, or for support, a Business License is arranged directly with us.

How do I access my files?

Native mounts on Mac, Windows, Linux, and Kubernetes (FUSE, NFS, SMB, FSKit, mountosio), plus the S3 and WebHDFS APIs the client itself serves. It is fully POSIX-compatible, so existing apps work as they are written.

How is POSIX compatibility verified?

With the industry filesystem suites, each covering a different surface. pjdfstest for POSIX compliance (8,789 tests), the Linux Test Project (LTP) for filesystem and system-call behavior, xfstests for filesystem regressions, fsx for I/O consistency under randomized operations, and IOR with mdtest for I/O throughput and metadata load.

Do I have to schedule backups or snapshots?

No. Version history at a configurable granularity (one minute by default, down to a sub-minute window), time travel, and recently-deleted recovery are built into the storage. Open any version or deleted file and copy it out.

Can it support analytical workloads?

Yes. An Iceberg Catalog volume type presents the data as an Apache Iceberg catalog, so analytical engines can query it directly.

Can I capture file events for processing?

Yes. Every general volume can keep an ordered feed of creates, deletes, modifies, and renames, and the mountos event subcommand streams it live or serves it as a local REST API.

Can AI agents work with it?

Yes. Each client ships a read-only MCP connector, alongside a drop-in agent skill and llms.txt for any MCP-capable client.