seedfast-mcp

Purpose

Seedfast aims to solve the problem of missing usable test data in development environments: it generates realistic and relationally valid synthetic test data based on the live schema of a PostgreSQL database and populates the database, so that teams that cannot put production data into development environments can still have an environment whose behavior is close to production.

Main capabilities

• Generate synthetic data based on the live schema, ensuring that data relationships are valid.
• Provide an MCP (Model Context Protocol) server as a public integration interface, including client configuration, tool reference, and smoke tests.
• Provide 13 tools, covering the complete loop of "planning a seed, running a seed, and checking execution status."
• Built in two prompts:seed-production-db(accepts a scope and a target database description) and scope-examples(converts a use case into an example scope).
• Already published in the official MCP registry, identified as st.seedfa/seedfast.

How to use

The MCP server is built into the Seedfast CLI and can be run directly via npx without separate installation. The configuration block can be placed in Claude Desktop's claude_desktop_config.json, Cursor's .cursor/mcp.json, or the corresponding configuration files for VS Code and Claude Code. For the paths for each client, see the MCP setup guide.

If you prefer a local binary, you can install the CLI and then point command to seedfast, with the argument set to ["mcp"].

The repository also provides a smoke test: it starts the published package via stdio, completes the MCP handshake, and lists the tools. This test does not require an API key because it does not access the backend. The expected output ends with the protocol version negotiated by the server and the number of tools it publishes.

Applicable scenarios

Suitable for teams that need a PostgreSQL database with behavior close to production in a development or test environment, but cannot introduce production data.

Limitations

• Currently only PostgreSQL is supported. Choosing PostgreSQL first is intentional, because its constraints and relational system are the richest, and that is exactly where the difficulty of generating trustworthy data lies.
• Support for MySQL, Oracle, and SQLite is under development.
• The repository contents are licensed under MIT; the Seedfast CLI is distributed under the same license, but its engine runs as a hosted service. The engine and the CLI itself are closed source.
• Regarding the specific details of configuration, tools, and prompts, as well as client paths, the material only points to the MCP setup guide and does not provide the full content; other unmentioned features, fees, prices, or effects are not provided for now.