Source: github.com/db-branch/db-branch
db-branch runs on Linux with ZFS and creates a branch of a base database by cloning a ZFS snapshot, so creation copies no data. Each branch runs its own database server on its own port. The project says that on its own three-node test cluster a PostgreSQL branch was ready in about 0.5 seconds and a MySQL branch in about 3 seconds.
Branches can be reset to their creation state, expire automatically with a time-to-live setting, or be saved with checkpoints and restored later. Per-branch upgrades move a single branch to a newer MySQL or PostgreSQL version, with a checkpoint saved first and automatic rollback on failure. Optional ProxySQL routing sends client connections to the right branch by user name or by a query comment.
The core is licensed under Apache-2.0. Authentication, audit, history, diff and FSx storage modules are closed-source paid packages enabled with a license key. The project's documentation says the paid modules run for 30 days without a key in its evaluation setup. Ubuntu 24.04 or 26.04 is required.
Why it matters if you build agents
Agents that run risky queries or migrations can be pointed at a disposable branch through the built-in MCP server, keeping production data out of reach of those steps.
CI pipelines can give each pull request its own database copy that is removed automatically by a time-to-live, instead of sharing one test database.
Because the project is in beta and its APIs and command-line details may still change before 1.0, pin versions before building automation on it.
Get stories like this every weekday
One short email with the day's agent tooling news.