DBSnapper

A command-line agent and cloud service that dumps a PostgreSQL or MySQL database, cuts it down to a related subset, masks sensitive columns with your SQL, and shares the result with developers as a snapshot they can load locally.

What it is: Database clients and design tools
Type: Commercial product
Made by: Scharfnado LLC
License: Proprietary (free standalone agent and Starter plan; paid Pro and Enterprise plans)
Facts checked: 23 October 2026

What it is

DBSnapper gives developers a copy of production data that is small and safe enough to work with. The agent, a single dbsnapper binary or Docker image, reads a YAML file of targets, each with a source database URL. dbsnapper build dumps the source with the database’s own tools (pg_dump for PostgreSQL, mysqldump for MySQL) into a compressed snapshot file; dbsnapper load drops and recreates a destination database and restores a snapshot into it. Only PostgreSQL and MySQL are supported.

Two steps sit between build and load. Subsetting copies only part of the data: you list tables to cut down with a where clause or a percentage of rows, tables to copy whole and tables to leave out, and the agent follows foreign keys so the rows those rows refer to come along. You can add relationships the schema doesn’t declare, or ignore ones it does. Sanitizing runs a SQL file of your own, such as UPDATE users SET email = …, against a restored copy and snapshots the result. That copy can be a database you name or an ephemeral one: a temporary Docker container on its own network that is removed when the job ends.

Without an account the agent runs in standalone mode with snapshots on local disk. DBSnapper Cloud, launched in January 2022, stores target definitions centrally with credentials encrypted, and writes snapshots to your own bucket on Amazon S3, Cloudflare R2, MinIO or DigitalOcean Spaces; teammates then find and download sanitized snapshots through presigned URLs, from the CLI, a VS Code extension or the API. Group-based sharing with Okta single sign-on, a Terraform provider and a GitHub Action round it out. The last release is a preview of version 3.0 from August 2025, which adds an MCP server so AI assistants can list, build, sanitize and load snapshots.

Why teams pick it

Subsets keep referential integrity: rows that a kept row points to come along, so the copy loads and the app works
Sanitization is plain SQL you write and review, run on a copy, and can run in a throwaway container
Raw data stays in your environment and snapshots go to your own bucket; the cloud service handles sharing and access
Uses the databases’ standard dump tools, and a free standalone mode needs no account

Watch out for

Only PostgreSQL and MySQL
Masking is only as good as your SQL: the docs describe no automatic detection of personal data, so a column the query misses ships unmasked
Closed source from a small company; team sharing, SSO, advanced subsetting and the Terraform provider need a paid plan
No release since the 3.0 preview of August 2025, which is still marked as a pre-release

Alternatives

What all database clients and design tools have in common →Compare side by side →

AtlasCommercialA schema-as-code tool, “Terraform for databases”: you describe the schema you want in SQL, HCL or ORM models, and Atlas plans, lints, tests and applies the migrations to get there.

ChartDBDormantOpen sourceAn open-source, browser-based ER diagram editor that builds a diagram from the JSON output of one query you run yourself, so it never needs a database password, and can export the schema as DDL for another database.

DB PilotCommercialA Mac database client that combines a table browser, a SQL editor and SQL-plus-Python notebooks with an AI assistant, and embeds DuckDB and chDB for querying local files.

dbdiagram.ioCommercialA web tool that draws entity-relationship diagrams from a schema you type in DBML, an open text language, and converts between DBML and SQL for PostgreSQL, MySQL, SQL Server and Oracle.

DbVisualizerCommercialA desktop SQL client written in Java that connects to almost any database through JDBC, with a free edition and a paid Pro edition that adds data editing, import and export, charts and an opt-in AI assistant.

How it fits with other technologies

What DBSnapper uses

A target with a subset and a sanitizing step (from the DBSnapper docs) · yaml

targets:
  sakila:
    name: sakila
    snapshot:
      src_url: mysql://root:mysql@localhost:13306/sakila?tls=false
      dst_url: mysql://root:mysql@localhost:3306/sakila_snap?tls=false
    subset:
      src_url: mysql://root:mysql@localhost:13306/sakila?tls=false
      dst_url: mysql://root:mysql@localhost:3306/sakila_subset?tls=false
      subset_tables:
        - table: sakila.film
          where: "film_id < 20"
        - table: sakila.actor
          percent: 20
      copy_tables:
        - sakila.store
      excluded_tables:
        - sakila.staff
      added_relationships:
        - fk_table: sakila.address
          fk_columns: city_id
          ref_table: sakila.city
          ref_columns: id
    sanitize:
      dst_url: mysql://root:mysql@localhost:3306/sakila_sanitized?tls=false
      query_file: sakila-sanitize.sql

See it in a real stack

No architecture uses DBSnapper yet. Drop it into the playground and wire it up yourself.Open playground →

Where it came from2021–2025

DBSnapper is the work of Joe Scharf, whose company Scharfnado LLC sells it. His argument was that hand-built test fixtures are “painstaking, time-consuming, and error prone” and go brittle, and that developers should work against sanitized copies of real data instead. It began as a command-line tool, first on GitHub in early 2021, where every setup kept its own local configuration file and snapshots stayed on local disk.

In January 2022 DBSnapper Cloud added central target definitions with encrypted credentials and storage profiles that put snapshots in your own cloud bucket; the standalone agent stayed free. Version 2.0, in February 2024, added subsetting, “smaller, relationally complete” snapshots, and switched PostgreSQL extraction to the COPY protocol. Okta single sign-on followed in June 2024, and a 3.0 preview with an MCP server for AI assistants in August 2025.

Why it’s built this way

Use each database’s own dump and restore tools: pg_dump and mysqldump already capture schema, indexes and data correctly, so the agent orchestrates them instead of writing its own export format.
Sanitize with your own SQL on a copy: Each schema stores sensitive data differently. Running a reviewed SQL file against a restored copy, then snapshotting that, lets teams mask exactly their columns without touching production.
Keep the data in the customer’s environment and bucket: Handing production data to a third party was the main worry. The agent processes data where it runs, and the cloud service only coordinates targets, access and presigned links to your own storage.

Timeline

2021
Joe Scharf publishes the first DBSnapper CLI on GitHub in January
2022
DBSnapper Cloud launches in January; agent 1.0 is released in November
2024
Version 2.0 adds subsetting in February; Okta SSO follows in June
2025
2.9 adds PostgreSQL schema filtering in July; a 3.0 preview adds an MCP server in August

Sources10 · facts checked 23 October 2026

Improve this page on GitHub