All lifecycle endpoints use the same token header:
Missing or invalid tokens return 401 Unauthorized. Keep the token in a secret store and avoid writing it to logs.
Start a run
Call the start endpoint immediately before doing the monitored work:
A start ping is accepted up to 60 seconds before the expected time. A repeated start ping is safe: while a run is already active, Still200 returns that same run instead of creating another one.
If a run was already marked missed, a later start may recover that missed occurrence. Its incident resolves when the start arrives, and a successful finish records the run as late.
Finish a run
Call the finish endpoint only after the work completes successfully:
The endpoint returns 409 Conflict when the monitor has no running job to finish.
Fail a run
If the work raises an error, report a concise message and optional structured diagnostics:
error_message is optional and limited to 2,000 characters. payload accepts a JSON object. If no message is supplied, Still200 records a generic failure message.
A process that exits without sending finish or fail remains running. Configure a maximum runtime if you want Still200 to turn that condition into a timed_out incident.
