[cbergh]

README only, no code. Opened as a draft for the product team to check before merging.

Why: search engines and AI assistants read this README to work out what TestGen is. It currently opens with "DataOps Data Quality TestGen is part of DataKitchen's Open Source Data Observability" and a run-on feature list, so answers describe it vaguely or skip it. The new first sentence is a plain definition that can be quoted as is.

What changed

• Name is "DataOps TestGen" throughout.
• Opening: one-sentence definition, then what the open-source edition includes (full web UI, CLI, MCP server, Apache 2.0, same core engine as Enterprise).
• "What it does" list; the three existing screenshots and the product tour stay.
• New "Supported databases": the 14 databases plus CSV, Parquet, and JSON on local paths, S3, or Azure Blob, as of 5.134.0.
• "Quick start" copied from the installer README (requirements, curl download, Windows .exe, tg install with --docker / --pip / --no-demo, tg upgrade, tg start). The two install-doc links stay.
• New "Use it from an AI assistant (MCP)" section linking the MCP docs, port 8530 / --api-port.
• New "Open source vs Enterprise" table, matching datakitchen.io/pricing (1 user, 1 connection, 1 project vs unlimited; $100 per user and per connection per month).
• Links: "Star us on GitHub" now points at this repo, certification link moved off info.datakitchen.io to /certifications/, product link to /products/dataops-testgen/, Twitter login-redirect URL replaced with x.com/datakitchen_io. All new links return 200.

Please verify before merging

1. Nothing is Enterprise-only beyond users, connections, projects, enterprise security, and support ("same core engine").
2. The testgen console script is a user-facing CLI; reword if not.
3. The MCP server is enabled in the open-source build with no license gate, on port 8530, and the docs URL is the right entry point.
4. "Every tool runs under your own TestGen role" matches testgen/mcp/permissions.py.
5. The 14-database list matches the connection flavors in code.
6. "Runs tests as SQL inside your database" is fair for file sources, which run on embedded DuckDB.
7. Installer flags --docker, --pip, --no-demo, --api-port against the current dk-installer.py.
8. Support row: Community Slack vs DataKitchen support.

Companion to #73 (PyPI metadata), which gives the package the same one-sentence summary.

[github-actions]

Coverage Report •

File	Stmts	Miss	Branch	BrPart	Cover	Missing
TOTAL	32855	22389	8926	20	25%

report-only-changed-files is enabled. No changed files were found in the coverage report :)

Tests	Skipped	Failures	Errors	Time
5	0 💤	0 ❌	5 🔥	40.518s ⏱️