Query big files without uploading them
Blinkql is a SQL workbench that runs entirely in your browser, for files too big for a spreadsheet and too sensitive to upload. It's being built in the open. An early version works today, but it's still slow to show large results.
SELECT machine_id, shift_date, SUM(run_min) / SUM(planned_min) AS availability, SUM(ideal_cycle_s * total_count) / 60 / SUM(run_min) AS performance, SUM(good_count) / SUM(total_count) AS quality, availability * performance * quality AS oeeFROM production_logWHERE shift_date >= '2026-10-01' -- this monthGROUP BY ALLORDER BY availability;| Row | machine_idVARCHAR | shift_dateDATE | availabilityDOUBLE | performanceDOUBLE | qualityDOUBLE | oeeDOUBLE |
|---|---|---|---|---|---|---|
| 1 | PRS-03 | 2026-10-06 | 0.612 | 0.874 | 0.951 | 0.509 |
| 2 | CNC-12 | 2026-10-05 | 0.688 | 0.902 | 0.968 | 0.601 |
| 3 | CNC-07 | 2026-10-07 | 0.701 | 0.913 | 0.982 | 0.628 |
| 4 | WLD-02 | 2026-10-06 | 0.734 | 0.887 | NULL | NULL |
| 5 | LTH-05 | 2026-10-05 | 0.756 | 0.921 | 0.973 | 0.677 |
| 6 | PRS-01 | 2026-10-07 | 0.781 | 0.896 | 0.977 | 0.684 |
| 7 | CNC-04 | 2026-10-06 | 0.802 | 0.934 | 0.985 | 0.738 |
| 8 | WLD-05 | 2026-10-07 | 0.823 | 0.908 | 0.989 | 0.739 |
For the exports a spreadsheet can't open
Machine logs, breakdown records, attendance sheets and stock counts often arrive as exports with a million rows or more. Spreadsheets struggle to open them, and what's inside is often too sensitive to upload. Blinkql opens the file in your browser tab and lets you ask questions of it in SQL.
- Queries run in your browser
- The engine runs inside the tab, on your computer. There's no server doing the work.
- Your files never leave your computer
- The browser reads the file directly. Nothing is uploaded, and there's no account.
- Built for exports of 1–2 million rows
- The size of a month of machine data from a plant, or a full stock ledger.
Work in progress
Blinkql is being built in the open. The work is split into eight milestones, each ending with a deployed app. The first three are done: the workbench opens a file of a million rows, queries it in your browser and stays responsive while it does. The current milestone, M3, Explore and visualize, adds column profiles, autocomplete, charts and export.
- M0FoundationsThe project scaffold, CI, deployments and the benchmark harness.Done
- M1Core query loopOpen a file, write SQL and see the results.Done
- M2Performance pass IQueries never block the page, even on a million rows.Done
- M3Explore and visualizeColumn profiles, autocomplete, charts and export.In progress
- M4Big files and persistenceFiles you keep reopen instantly; remote files by URL.Planned
- M5Operations packFactory sample data and templates such as OEE and downtime Pareto.Planned
- M6Performance pass IIA faster first load, offline use and a canvas results grid.Planned
- M7Version 1.0Accessibility, shortcuts, shareable query links and full docs.Planned
Try the early workbench, see how fast it is today, read the roadmap or follow the milestones on GitHub.
Help build it
You don't need to write code to make a difference. Blinkql is shaped by the people who have the files: engineers, analysts and supervisors who know what the data looks like.
- Run the benchmarks on your hardware
- Run the suite on your own laptop and share the numbers, so performance claims rest on more than one machine. How to run the benchmarks
- Share datasets and templates from your industry
- Synthetic files and analysis templates from manufacturing, logistics, retail, finance or healthcare operations. Never real data.
- Review the design
- Check screens and flows against the Design Language and suggest improvements.
- Review accessibility
- Try it with a keyboard or a screen reader, and report what's hard to use.
- Improve the documentation
- Fix unclear pages in the wiki, add examples or write a tutorial.
- Write code
- Pick a small, well-scoped issue that comes with guidance, and open a pull request.