Endpoint
One link, any feed.
Paste a public URL or a private one. Donum talks to it from the workspace, not from the browser.
Live Data
Pull on demand. Watch the page redraw.
The shape
Point at any feed. The parser picks the format from the response and lays the rows in.
// CSV, the simplest case.GET https://api.example.com/v1/q4.csvRegion,Q1,Q2,Q3,Q4North,12400,14800,16200,18900// JSON, three shapes accepted.[ { region: "North", q1: 12400, ... } ]{ data: [ ... ] }{ rows: [ ... ] }
Auto detect
Content type guides the choice. CSV, TSV and JSON each land as a typed grid with column inference.
Three knobs
A URL, an optional header, and a click. The page stays current.
Endpoint
Paste a public URL or a private one. Donum talks to it from the workspace, not from the browser.
Auth
Bearer tokens, API keys, custom headers. They live on the server. The frontend never receives the secret.
since last pull
Cadence
Click Refresh whenever you want fresh rows. Scheduled cadences land in a future phase, sitting on the same engine.
The contract
20s
timeout per pull
5MB
payload cap
0
tokens in the browser
A workspace stays responsive when the worst feed misbehaves. The cap, the timeout, and the parse errors return clear codes. The grid keeps showing the last good rows.
In short
Anything that returns CSV, TSV, or JSON. Auto detection picks the parser. JSON arrays, data wrapped, and rows wrapped all land cleanly.
On the server. Donum makes the call from the workspace, not the browser. The token never reaches the client.
Five megabytes per pull and a twenty second timeout. Both keep the workspace responsive.
Yes. Every chart bound to the source updates the moment the new rows arrive.