You can now tell Depot CI ahead of time that a step might need a retry. Add retry to a known-flaky step and Depot reruns just that step, not the expensive work before it.
The Depot CI API and CLI are now generally available, so everything you could do in the dashboard you can now do from your terminal or hand off to an agent.
The longer I worked on a problem with a coding agent, the dumber it felt. My context window was a junk drawer. Here's the multi-role loop I built so plans, builds, and reviews each get their own clean context.
The old CI debugging loop goes: edit YAML, commit, push, wait, squint at logs, repeat. Depot CI replaces that with a local loop you can actually control, from running workflows against uncommitted changes to SSHing into the runner mid-job.
Offload the scavenger-hunt grunt work to agents, log the useful bits (also with an agent) so the problem is discoverable next time, and spit out the day's support recap in seconds.
I used an LLM to refactor a 1500-line mess into clean, organized code. I realized months later that agents were the only ones working in it. They didn't need the clean code. They didn't need my empathy. My intuitions as an engineer just don't work anymore.
I pegged every CPU in our ClickHouse cluster during my second week. The fix was quick, but it triggered the three fears that actually make engineering suck: looking dumb, underperforming, and wasting people's time.
