Browse, Query, Chart.

A desktop app for exploring production databases. Map the schema and query a sample. Pivot it, chart it. Then export the whole thing as an interactive report.

Download for Windows Read the docs

Signed macOS installer coming soon · Apache-2.0 · Unapologetically not Excel

The safe-db query builder with a joined, filtered query and a 40-row result sample

Drag, drop, connect. Build complex queries without the SQL.

See what you’re querying

The schema map lays out tables, keys, indexes, and referenced external tables, so you can see how the data fits together before you ask it anything. Most SQL clients assume you have the schema memorized. safe-db draws you a map.

Queries are validated against the loaded schema and compiled with bound parameters, so what runs is exactly what you built, aimed at columns that actually exist.

safe-db schema map showing tables, keys, and relationships

safe-db SQL screen with a single SELECT statement and schema-aware autocomplete

SQL, within reason

The SQL screen takes a single SELECT in your database’s dialect, with schema-aware autocomplete. It’s parsed into the same structured, parameter-bound query the builder produces. Your text never reaches the database. Writes, subqueries, and aggregates are rejected with an explanation.

Bounded by default

“Download everything” is not a feature. Every run is scored against an adjustable risk profile. A missing plan or a missing optimizer cost needs an explicit confirmed retry.

Default sample: 500 rows. Enough to answer the question.
Hard cap: 10,000 rows. The ceiling is not negotiable from inside the app.
Query timeout: 10 seconds, then the database is left in peace.
Advisory: A nudge above 1,000 rows, in case the question got ambitious.

Explore the sample, not the server

Pivot, Worksheet, and Visualization all work from an immutable sample. Tinker with the view as long as you like. Then export it to CSV or HTML.

safe-db pivot view aggregating the query sample — More comfortable with a pivot table?

safe-db visualization view with a bar chart built from the query sample — Or is a chart more your speed?

Reporting, reimagined

Most exported reports are a screenshot with ambitions. Export HTML writes the whole exploration into a single self-contained file: tables sort and filter, the chart renders inline, and clicking a mark drills into the same source rows the app would show. Whoever opens it gets the interactivity without a database connection, credentials, or a live query to re-run. CSV and PNG exports are still there, for the traditionalists.

An exported safe-db HTML report with a revenue-by-order-status bar chart and its sortable, filterable data table

The recording is the actual exported file, opened in a browser. Sort, filter, and drill into the rows; no app, no connection, no setup.

Credentials stay in the keychain

Named connection profiles are saved without passwords. The OS credential store (Keychain on macOS, Credential Manager on Windows) holds secrets when it’s available. Sticky notes are not a backend.

safe-db connections screen listing saved connection profiles

Four engines, one experience

PostgreSQL

Verified TLS with pgjdbc’s normal trust behavior.

MySQL

JVM or launch-profile trust stores.

SQL Server

JVM or launch-profile trust stores.

Oracle

Verified TCPS via an Oracle wallet.

Give them a database client you don’t have to worry about

Signed Windows installer today. The macOS app already runs from source; a signed installer is next. If you’ve ever been paged because someone “just needed a count,” you already understand the product.

Get safe-db