Datasets

Upload a file of examples, or open a dataset that ships with the install. Both give you a draft. Nothing runs until the draft is started.

Endpoints

Under /api/v1
Method Path What it does
GET /datasets Every dataset this identity may read: its own, then the ones that ship with the install.
GET /datasets/{name}/pipeline One dataset’s declared pipeline.
GET /datasets/{name}/cells The measured cells of a dataset.
GET /datasets/{name}/cells/{answer} One answer opened: its row assembled into a trace.
POST /datasets/ingest Parse an uploaded tabular file, mint a check-in campaign and return its draft.
POST /datasets/{name}/draft Open an authored dataset’s files as a prefilled check-in campaign.
POST /datasets/draft/candidate-library Attach an uploaded candidate library to a draft and return the updated draft.
POST /datasets/draft/candidate-library/from-column Build a draft’s candidate library from the distinct values of one of its own columns.
GET /origins Every runnable origin in the caller’s tenant, newest first.
POST /origins/{origin_id}/draft Open a prior origin as a prefilled check-in campaign. Nothing runs until start-checkin.

POST /datasets/ingest

A multipart upload. file is CSV, TSV, JSON, JSONL or XLSX; slug is optional. The answer is the draft of a check-in campaign. Edit it with the edit-draft-campaign command and start it with start-checkin.

curl -X POST http://127.0.0.1:8001/api/v1/datasets/ingest \
  -F "file=@tickets.csv"

Cells

/datasets/{name}/cells takes limit and scope, which narrows the cells to the dataset, a campaign or a cycle.

See also

Commands
start-checkin, edit-draft-campaign, resolve-origin.