AutoPilot for CI savings
Measured savings of jobs AutoPilot for CI routed off their default runner, in total and per workflow. Each workflow AutoPilot made a decision for in the period is listed, also with no savings yet, with the share of its jobs that ran on each runner (largest first), its AutoPilot state (managed when it routed a job off the default runner, learning when every job lacked history, otherwise limited with the most common reason), the settings set in the workflow file (locked_keys) and the effective policy with where it comes from. Each job is compared with the p75 cost and wall time of the same job on its default runner over the 30 days before AutoPilot first routed it. When the job had fewer than 10 runs there, the baseline is estimated as the default runner’s list price for the job’s own measured run, with no time saving claimed (estimated_jobs); only when that is unknown too does the job count in no_baseline_jobs and add no savings. A job that would have run free on its default runner saves nothing. net_usd subtracts what the routing jobs themselves cost. Defaults to the last 30 days.
Authorizations
Enter "Bearer {token}"
Headers
Tenant ID
Query Parameters
First day, YYYY-MM-DD
Day after the last day, YYYY-MM-DD
owner/name
Workflow file
Runner provider the jobs were routed to