Two ways to use it

Same cleaning rules in both. Pick the surface that matches how you work.

In your browser

No uploadNo serverFree

Open the page, drop a workbook in, and the cleaning runs locally on your machine. The file never leaves the tab. Best for one-off cleaning, demos, and privacy-sensitive workflows.

Try the in-browser tool

As a server

ExpressAPI key authRate limitedVirus-scan hook

Run a Node process behind a POST /upload endpoint. Same cleaning, plus optional ClamAV scanning, Redis-backed rate limits, configurable retention, and downloadable per-sheet outputs.

See the API reference

How it works

The pipeline is three small modules. Each does one thing, and they hand each other plain JavaScript values.

Extract

Read every sheet from the uploaded workbook into a 2-D array of cells. Empty sheets are skipped.

Transform

Apply column rules (NHI → ID, DOB → Age, drop Address/Contact), skip blank rows, and dedupe. Per-sheet stats are emitted alongside the cleaned rows.

Load

Write one CSV per sheet plus a per-upload manifest. The route returns JSON pointing at every output file by name.

One curl away

The HTTP surface is small on purpose. One endpoint, one response shape, no surprises.

$ curl -F "excel=@./case-mix.xlsx" \
       -H "Authorization: Bearer $CLINISYNC_API_KEY" \
       http://localhost:3000/upload

{
  "message": "Upload and transformation completed successfully.",
  "sheetsProcessed": 3,
  "rowsProcessed": 17,
  "duplicatesRemoved": 1,
  "invalidDobCount": 1,
  "missingNhiCount": 1,
  "files": [...],
  "manifest": "manifest-case_mix-...-.json"
}

Why Thresh

Three principles the codebase actively defends.

De-identify by default

Output never contains the original NHI, DOB, address, or contact details. The most common mistake (forgetting to scrub a column) is prevented by the pipeline, not procedure.

Hold no data

Cleaned files are swept by a TTL sweeper. There is no database, no patient registry, no audit table — just transient on-disk artifacts and a per-upload manifest.

Honest about scope

Thresh isn't HIPAA-certified, isn't a managed service, and isn't a place to store data. It is the cleaning layer that sits in front of one — and the docs say exactly that.

Drop in a workbook?

The in-browser tool processes the file locally. Nothing is uploaded.

Open the in-browser tool Read the docs