Features

Everything Seedfast does, grouped by area. Because a capability list is worth only as much as it can be verified, each row links to the documentation page that covers it.

Schema handling

What Seedfast reads before generating anything, and what the database accepts afterwards.

Live schema reading
Seedfast reads tables, columns, types and relationships on every run, so nothing needs redeclaring after a migration. Docs
Referential integrity
Every foreign key in the generated data points at a row that is already there. Docs
Constraint-valid values
Enums, check constraints and unique columns receive values they permit. Docs
Trigger-aware inserts
Triggers that fire on insert are accounted for, so the stored result matches how the application itself writes. Docs
No configuration file
A connection string is the whole setup. There is no schema definition or per-column rule file to keep. Docs

Data generation

How the values get chosen once the schema has been read.

Domain inference
Table and column names tell Seedfast what the data represents before any value is generated. Docs
Language and format control
Ask in the scope for a language or a particular column shape, and the values come back that way. Docs
Scope selection
You pick which tables get filled and how many records each one receives. Docs
Append-only writes
Rows already sitting in the scoped tables are left alone, because Seedfast adds records rather than replacing them. Docs

Volume

One workflow covers a local branch and a load test.

Large-volume seeding
Fifty records for a local branch, or tens of millions when the tables need to weigh what production weighs. Docs
Realism at scale
Value quality does not degrade as the record count climbs into the millions. Docs

Run control

A run tells you what it will do before it does it.

Seeding plan preview
Per-table record counts are shown for approval before anything is written. Docs
Scope validation
A scope naming a table that is missing gets caught before the first insert. Docs
Machine-readable output
Alongside the default terminal view, a plain mode suits CI logs and a JSON mode suits scripts that parse the result. Docs
Documented exit codes
Distinct codes cover authentication failure, database connection failure, quota exhaustion and cancellation. Docs

Where it runs

Seedfast is reachable from a terminal, a pipeline, or an AI assistant.

Command line
One command runs against a connection string, whether the database sits locally or remotely. Docs
CI/CD pipelines
Seedfast runs as a pipeline step, with GitHub Actions, GitLab CI and CircleCI documented. Docs
MCP server
Assistants such as Claude and Cursor seed a database through a tool call. Docs
Non-interactive authentication
Local runs use a browser login, and pipelines authenticate with an API key instead. Docs
Cross-platform install
A single binary covers macOS, Windows and Linux, installed through Homebrew or npm. Docs

Data handling

The run sends schema metadata and nothing else.

Generated from scratch
Production records are never used as source material or as training input. Docs
Credentials stay local
Database credentials remain on your machine, and only schema metadata is transmitted. Docs

Database support

Engine coverage, stated as it stands today.

PostgreSQL
PostgreSQL is supported today. Support for other SQL engines, including MySQL, Oracle and SQLite, is in development. Docs

Stop using production data

Generate realistic synthetic data from your database schema.Safe, coherent, and ready for local dev, demos, and CI.

Get started for free