anybranch
Branch your database like code. An isolated, writable copy of production in seconds — any size, any engine, on infrastructure you own. Open source, self-hosted, Apache-2.0.
anybranch keeps a live replica of your production database on a server you run, then cuts copy-on-write branches off it. Each branch is a real, writable database with its own port and its own generated credentials. A 50 GiB branch costs the same as a 50 MiB one, because nothing is copied until a page changes.
What it does
- Replica once, branch forever. One initial copy of production; every branch after that is instant.
- Any engine. Postgres, MySQL, MongoDB, and SQLite, branched the same way.
- Zero blast radius. Branches are sealed copies with their own credentials — a leaked branch URL can’t reach production.
- Branch from any branch. Fork a branch of a branch, like git.
- Prepared pools. Pre-warm snapshots so an agent can claim a ready branch in well under a second.
How it compares to hosted branching
Managed branching services sell the same idea — a replica of production plus instant branches — as a Postgres-only, multi-tenant cloud product. anybranch is the open-source, self-hosted shape of that: you deploy the server, it uses each engine’s own replication instead of a proprietary pipeline, it covers four engines, and your data never leaves your machines. It has no accounts, orgs, billing, or bring-your-own-cloud — it is one server with one admin token.
Explore
- Quickstart and Install / Server deployment
- Connect a database: overview, Postgres, RDS, Supabase, MySQL, MongoDB
- Concepts: Architecture, Security
- Reference: CLI, Configuration, REST API, FAQ
- Workflows: AI agents, CI/CD, Local development