Runs

Start runs, follow their status and logs, abort, resurrect or reboot them.

📄️Start a run

Start a run; `webhooks` apply to this run only; `wait_for_finish` (0–60 s) answers once it finished or the time is up

📄️Runs, newest first

Runs, newest first

📄️One run

One run (also with the run's own token)

📄️The actor's status message

The actor's status message (run token or owner): {status_message, is_status_message_terminal}; a terminal one is kept after the run ends

📄️Pay-per-event charge while the run is active

Pay-per-event charge while the run is active (run token or owner): {event_name, count}; event_name follows the actor pricing rule (letters, digits, then also . _ -, up to 60) and may not be result_item (metered from the result count); answers { charged_event_counts }

📄️The run becomes a run of another actor

The run becomes a run of another actor (public or the owner's) on the same storages: {target_actor_id, input, build?}; the input is stored under a new INPUT-METAMORPH-n key

📄️Add login session to run

Run token only: a social actor run borrows a login pool account {platform}; answers { id, platform, username, cookies, user_agent, proxy { groups, country, session }, lease_expires_at }

📄️Report login session

Run token only: how the borrowed account did {outcome: ok|error|rate_limited|logged_out|checkpoint|banned|released, message?}

📄️List views (runs)

Dataset views and outputs of the actor version the run used: `views[]` ({name, title, transformation, display, query}) where `query` is the dataset items parameters (fields, unwind, flatten, omit, desc) of the view; `outputs[]` ({name, title, url}) from the output schema

📄️Bulk action on runs

One action on up to 200 of your runs, each as its single endpoint does it: abort (a READY run is aborted at once, a RUNNING one stops on its runner's next heartbeat) or resurrect (on its own storages and options; limits, balance and concurrency are checked per run). A run that cannot take it (abort: already finished; resurrect: still active, storages gone, refused) is listed in `failed` and the rest go on. Ids that are not your runs or repeated: 422, nothing changed. 20 a minute

📄️Abort a READY or RUNNING run

Abort a READY or RUNNING run

📄️Start a finished run again on the same storages

Start a finished run again on the same storages; `options` {memory_mb, timeout_secs, build} may change; 422 when it is not finished

📄️Restart the container of a RUNNING run

Restart the container of a RUNNING run (it keeps its storages and state); 422 otherwise

📄️Log tail

Log tail (secrets masked), at most 1 MB per call (log_page_bytes); ask again with from=offset until finished and an empty chunk. A job keeps at most 50 MB of log, then a truncation marker