Documentation
Build with SynthForge
Guides for automating synthetic data from scripts, CI, and local tools. Prefer the visual app? Launch the app and generate from the UI.
API keys & automation
Create a personal API key in Settings, authenticate with Bearer tokens, and generate datasets from curl, Python, or the CLI. First key includes 250 free credits.
Hosted MCP
Point Claude, Cursor or any Streamable HTTP client at mcp.synthforge.io and generate from one prompt. First API key includes 250 free credits; buy more any time from the MCP page.
Python
Create a schema, start a job, wait, download, in about forty lines of requests.
CLI
The synthforge command for terminals and CI.
TypeScript SDK
Typed client for the REST API with a polling helper.
PostgreSQL / Neon / Supabase
Generate multi-table Postgres fixtures with real FKs and load with psql (local, Neon, or Supabase).
MySQL / MariaDB
Generate multi-table MySQL fixtures with real FKs and load with mysql --local-infile.
DuckDB integration
Generate multi-table DuckDB fixtures with foreign keys and load them with one COPY script.
SQL Server
Generate multi-table SQL Server fixtures with real FKs; fill in the CSV directory, then BULK INSERT via sqlcmd.
CockroachDB
Generate multi-table Cockroach fixtures; userfile upload then IMPORT INTO with import_cockroachdb.sql.
Benchmarks vs CTGAN
Open methodology and numbers: schema-driven generation vs a learned baseline. Honest tradeoffs.