What it does
Build-test-deploy engineering on data you supply: finds where a CI run spends its wall clock and what to change, fits a trend to your run history to name the stage that will bottleneck first, turns a change and your deployment record into a staged rollout with gates, works out the point past which a rollback is no longer a redeploy, and writes the runbook for the night it goes wrong.
Example prompts
- Here are the step timings from our last CI run — where is the wall clock going, and what would you change first?
- This is our run history for the past month. Which stage will become the bottleneck, and is it doing more work or waiting longer for a runner?
- We're shipping a schema split through staging, canary and production. Plan the rollout, and tell me the gate on each stage.
