Install open-source DataOps TestGen
Install open source data quality and observability in minutes.
Never write a data test again. TestGen profiles your warehouse, generates thousands of tests, and flags 27 classes of issues on the first run. Full UI, runs unlimited in your environment.
# TestGen install for Mac, Linux.
# Download the latest installer.
$ curl -o dk-installer.py \
'https://raw.githubusercontent.com/DataKitchen/data-observability-installer/main/dk-installer.py'
# Run the install command.
$ python3 dk-installer.py tg install
# Full docs: docs.datakitchen.io/testgen/get-started/install-on-mac-linux/# TestGen install for Windows.
# Download the latest installer.
PS> Invoke-WebRequest `
-Uri 'https://github.com/DataKitchen/data-observability-installer/releases/download/latest/dk-installer.exe' `
-OutFile dk-installer.exe
# Run the installer.
PS> .\dk-installer.exe
# Full docs: docs.datakitchen.io/testgen/get-started/install-on-windows/- Apache 2.0
- Runs on your infrastructure
- No data leaves your environment
See it first
Take a tour before
you install.
Click through TestGen's profiling, hygiene detection, and auto-generated tests on a real dataset. No signup. No install.
Found a data error?
Here's what TestGen flags on the first profiling pass.
Point it at your warehouse. Twenty-eight minutes later, the issue list is in your browser. Like these:
- Null-rate spikes
- Stale timestamps
- Broken foreign keys
- Schema drift
- Uniqueness violations
- Row-volume spikes
- Outlier values
- Missing required values
Plus 19 more. Out-of-the-box Freshness/Volume/Schema Anomaly Detection.
What you get, out of the box.
Everything below is in the open source release. No feature gating, no upgrade prompt.
51 profiling characteristics
Profile every column in every table: data types, null rates, cardinality, distributions, patterns. The foundation everything else builds on.
27 data hygiene detectors
Find the things that quietly break downstream: stale timestamps, broken keys, unexpected encodings, columns drifting out of spec.
Thousands of auto-generated tests
Every profile becomes an active test. Schedule them, score the results, surface what changed. No hand-written assertions to maintain.
In-database execution
Tests run where the data lives. Snowflake, Databricks, Postgres, BigQuery, MS SQL, and more. No bulk extract, no second copy of your warehouse.
Built-in data catalog
Every table, every column, every test. Searchable. Linked to scores and history. Your team stops asking "what's in this table again?"
Anomaly detection
Track scores over time. TestGen flags the runs that drifted, not just the ones that failed an absolute threshold.
TestGen Enterprise
100% of the features.
10% of the price.
Two Apache 2.0 tools that profile your data, detect anomalies, monitor pipelines, and catch issues before they reach your customers. Built by the team that pioneered DataOps.
Compare Pricing- Unlimited testing, flat price. Test every table in every database. No per-table fees, no credit meters.
- Open source core, Apache 2.0 licensed. Start free with no feature gating. Profiling, hygiene checks, anomaly detection — all in.
- No vendor lock-in. Self-host on your own infrastructure. Your data never leaves your environment.
- Bootstrapped since 2013. Profitable and independent. Our success depends on yours, not investor demands.
Honest comparison
We built TestGen because writing data tests sucks.
hard pass
Great Expectations and Soda Core are excellent frameworks for hand-writing data quality tests. TestGen learns your data and generates the data tests you need in minutes. You get a UI to share tests with your customers, an MCP server, and a library of custom best practice tests.
| Capability | Next-generation open source DataKitchen TestGen | Previous generation open source Soda Core · Great Expectations |
|---|---|---|
| Web UI | Built-in | CLI + Python or YAML |
| Test authoring | Automatic | Enjoy spending hours writing Python per expectation or YAML per check |
| Data profiling | 51 characteristics per column | Basic stats only |
| Hygiene detection | 27 detectors out of the box | Manual |
| Built-in data catalog | Yes | No |
| Anomaly scoring over time | Yes | No |
| MCP support | Full MCP interface | No |
| Self-hosted, runs inside your firewall | Yes | Yes |
All three projects are Apache 2.0 and run inside your environment. The difference is what you have to build yourself.
Ready? Scroll up and install. Really. Or check out our github repo.