Why this exists.

Every failing CI build starts with the same question, and it is not "what's wrong with my code?" It is: which subsystem broke. Did the runner run out of memory, did the registry time out, did a lockfile drift, is the test just flaky? Answering that means recognizing the same log patterns, over and over, usually at the worst possible moment.

PatchRail is that recognition written down as code. A CLI that matches a raw CI log against 31 failure classes using local rules, redacts secrets before it writes anything, and points at a step-by-step fix guide. It runs entirely on your machine: no account, no telemetry, no network calls.

It is Apache-2.0, and the open-source tool is the whole product — there is no paid tier behind it. The 31 fix playbooks are published free on this site, and the classification rules live in the repository where anyone can read or correct them.

If PatchRail misreads one of your logs, or a class you keep hitting is missing, open an issue with the redacted log. That is how the rules get better. And if it saves you a debugging morning, a star helps other maintainers find it.