Database changes that glide to production.
Plan, review, approve and promote schema changes across 17 engines, with rollback ready at every step and credentials that never leave your network.
MySQL
MariaDB
BigQuery
Databricks
Redshift
ClickHouse
Cassandra
CockroachDB
TiDB
YugabyteDB
DuckDB
SQLiteEvery step of a change, in one place.
Explore, assess, compare and promote without jumping between scripts, tickets and direct database access.
Every tool. One calm workspace.
Explorer
Browse, query and diagram every connected engine under per-environment roles, with production read-only by default.
Compare & Sync
Schema and data diff between two environments on the same engine, with an order-safe script that closes the gap.
GitOps
SQL change scripts in GitHub or GitLab, synced on merge and mapped to your promotion flows.
Conversion Studio
Analyze a source, draft the target model, review overrides and run one-time or CDC change scripts.
Drift Monitor
Scheduled snapshots against a baseline, so environments stop quietly diverging.
AI and MCP
Claude Desktop, Cursor and Codex inspect workspaces and review risk, with confirmation gates.
Audit ledger
Hash-chained, verifiable history of who did what, and when.
17 engines
From PostgreSQL and SQL Server to Snowflake and MongoDB. Capability depth varies by engine.
The cloud orchestrates. Your Executor holds the keys.
Pipelines, approvals and audit live in the control plane. The Executor inside your network holds every connection string and executes the change.
- Per-environment RBAC and multi-party approvals
- Query results pass through without row retention
- Separation of duties between authors and approvers
Your AI agent can propose.
A person approves.
Executors end up with more access than anyone intended a token pasted in to unblock a task, a role granted for one afternoon and never taken back. On the environments you protect, a change still waits for a named person before it runs.
Approval is a human decision
Production waits for the approvals you require, each recorded against the person who gave it.
Never more access than the person
An assistant acts with the signed-in user's roles per environment. Read-only stays read-only.
Every action is attributable
What ran, on which environment, and who was behind it, in a hash-chained ledger you can verify.
Common questions
Can we lock down production?
Yes. Use per-environment RBAC and multi-party approvals so production stays tightly controlled.
Which databases are supported?
17 engines: PostgreSQL, MySQL, MariaDB, SQL Server, Oracle, MongoDB, Snowflake, BigQuery, Databricks, Redshift, ClickHouse, Cassandra, CockroachDB, TiDB, YugabyteDB, DuckDB, and SQLite. All of them are explored, compared, governed and promoted the same way. Where an engine's own design rules something out, it is listed on the supported databases page.
How does GitOps fit in?
Migratrix can sync SQL-based change scripts with GitHub or GitLab and map them to promotion flows.
Where do our database credentials live?
On the Executor you run inside your network. See the trust center for architecture and data handling.