Publisher run list (calls to my apps)
Every call made to the apps I publish, most recent first — the
detail face behind GET /publisher/usage: the same ownership (“apps I
own”, zero-Release apps included), the same range predicate
(created_from / created_to are ISO-8601 absolute timestamps,
inclusive start / exclusive end, by run start time) and the same
filter values, so any aggregate number drills down to exactly the
runs it counted.
This is a whitelisted view, not the caller’s run object: it tells
you what happened (state, timing, records written back, error code
and message, data fee the caller paid) and never who called, what
they sent (input) or what they got back (no dataset_id, no
records, no credential identifier). Callers keep reading their own
runs through GET /runs; a publisher cannot open one of these run_ids
there. Debug traffic (run_kind=test) is excluded unless asked for;
status accepts comma-separated states (OR). Unknown or other
people’s apps in data_app are 404 (no existence leak).
Authorizations
DataHub API Key
Query Parameters
restrict to these apps (<username>/<app_name>, repeatable); default = all my apps
run state, comma-separated multi-value OR (e.g. FAILED or PENDING,QUEUED); values outside the vocabulary are 400
restrict to one calling channel (api / sdk / mcp / web …, same value as the runs list filter)
restrict to runs pinned to this Release version
restrict to runs whose error code equals this (the error_code key of GET /publisher/usage group_by=error)
production (default) live traffic only, test my debug runs only, all both; platform health probes never appear; other values are 400 invalid-run-kind
x >= 01 <= x <= 200Response
Successful Response
Page of calls to my apps, most recent first.