Early Access

Recreate any test state
in seconds.

Generate state-based test data and reset your database for E2E tests, API tests, and local development.

~/myapp — seedmancer

# scenario: billing/pro — premium plan users

The problem

Test data is harder than it looks

Every one of these slows your team down. Seedmancer solves them all.

Creating test data manually is painful

Every new test scenario means manually inserting rows, setting up relationships, and keeping it all consistent. It takes hours, breaks easily, and nobody wants to maintain it.

Schema changes quietly break your test data

Add a column, rename a field, change a constraint — and your saved test data silently becomes invalid. You only notice when tests start failing in ways that make no sense.

Live data gets dirty. Resetting to a known state is hard

After a few test runs, your database is full of noise. Getting back to a clean, predictable state means dropping tables or running fragile scripts — every single time.

Who it's for

Built for QA engineers and developers

QA / Test automation engineers

Run every E2E and API test from a predictable database state. Create state-based test data, reset state between tests, and reduce flaky failures caused by messy or shared data.

Predictable starting state for every test run
Named states per test suite or feature
No shared data causing unexpected failures

Developers

Reproduce bugs and test features locally without manually editing your database. Create users, orders, subscriptions, permissions, and edge-case data for the exact state you need.

Instantly recreate any app state locally
Edge cases and permission edge cases on demand
Stop manually seeding data between features

How it works

From state to reset in three steps

Export a snapshot

Capture your current schema and data as a versioned snapshot you can reset to anytime.

$ seedmancer export myapp/baseline

Generate test data with AIPro

Describe the state you need. The AI writes the SQL, runs it, and saves it as a new revision. When your schema changes later, seedmancer refresh adapts the data automatically.

$ seedmancer generate myapp/edge-cases --prompt "users with expired subscriptions"

Reset to any state

Restore the exact database state in under a second — before tests, during debugging, or for local development.

$ seedmancer seed myapp/edge-cases

Get started

Try it in 30 seconds

Install the CLI, connect your database, and export your first snapshot — free.

$ go install github.com/KazanKK/seedmancer@latest

Features

Everything you need for reliable test states

A focused CLI that handles state materialization, state reset, and cloud push and pull — so your tests stay reliable.

State-based datasets

Define named states like "premium users with orders" or "users for sorting tests" and generate the exact data you need every time.

Instant state reset

Restore your database to a known state in under a second. Run before every test suite so each run starts from the same baseline.

Cloud push & pull

Push and pull dataset revisions across environments. Share consistent test data with your team without copying files.

CI/CD ready

Add seedmancer seed before your test command. Every CI run starts from the same predictable state.

Built-in MCP server

Drive Seedmancer from Cursor, Claude Desktop, and other MCP hosts. Agents get typed tools — no stdout parsing.

AI Agents · MCP Server

Let your agent manage test state

Seedmancer ships a built-in Model Context Protocol server. Plug it into Cursor, Claude Desktop, Continue, or Zed and your AI agent can export, seed, generate locally, and reset state through typed tool calls — no stdout parsing, no shell hacks.

Typed tools for every command: seed_database, generate_dataset_local, push_dataset, pull_dataset, and more.
Structured resources for datasets, schemas, and live status — agents browse without guessing.
Safety rails built-in: destructive tools carry the MCP destructiveHint and prod-like env names are refused by default.

.cursor/mcp.json

{
  "mcpServers": {
    "seedmancer": {
      "command": "seedmancer",
      "args": ["mcp"]
    }
  }
}

Then ask your agent

“Reset my database to the premium-users state before running Playwright.”

Simple pricing, no surprises

Start free. Upgrade when your tests outgrow the limits.

MonthlyYearly

Free

$0/month

Snapshot create & restore
Unlimited local AI generation via MCP
MCP server for AI agent integration
Unlimited scenarios & revisions on your machine

Recommended

Pro

$5/month

Everything in Free
AI test data generation (seedmancer generate)
Schema-drift refresh (seedmancer refresh)
Cloud push & pull (up to 200 dataset versions)
Priority support

FAQ

Frequently asked questions

Which databases are supported?

PostgreSQL and MySQL are both fully supported. Point Seedmancer at a mysql:// or postgres:// URL and the same export, seed, and generate workflow works unchanged.

Does this work with Cypress and Playwright?

Yes. Add seedmancer seed myapp/api-test --yes before your test command and each run starts from a predictable, consistent state.

How is this different from a seed script?

Seed scripts are static — they're hard to maintain and break as your schema evolves. Seedmancer stores state-based datasets that are versioned, sharable, and can be regenerated against your live schema at any time.

Can I generate test data locally?

Yes, and it's free on every plan. Use seedmancer generate-local with a full SQL script (TRUNCATE + INSERT) piped on stdin, or drive it from an MCP agent. The entire run stays on your machine.

Which features need a Pro plan?

Cloud AI generation (seedmancer generate), schema-drift refresh (seedmancer refresh), and cloud push & pull. Export, seed, and local AI generation through the MCP server are free.

Does it work with Cursor and Claude Desktop?

Yes. Seedmancer ships a built-in MCP (Model Context Protocol) server. Point Cursor, Claude Desktop, or any MCP host at `seedmancer mcp` and the agent can export, seed, generate, and push or pull scenarios through typed tool calls.

What is coming next?

Dataset diffing, branching, and hosted team workflows are planned for upcoming releases.