Repo Feature Planner
Explores your repository and returns a commit sized plan, the risks and a definition of done before it writes code.
You fill in describe the feature you want…
A Claude Code prompt that reproduces a failing build locally, names the real cause and fixes it without skipping or deleting a single test.
Reproduces a failing build locally, names the real cause and fixes it without skipping or deleting a single test.
I want you to act as the engineer on call for a broken build. I will paste the failing job's log between <log> and </log> tags and tell you which branch it is on. Start by reading the log to the end and quoting the first real error, not the last line printed. Then reproduce it locally: find the command the pipeline runs, run it yourself, and show the output. If it passes locally, say so and investigate the difference between the two environments, such as tool versions, environment variables, caches, file ordering or timezone, before you change any code. Tell me which of these the failure is before fixing anything: a genuine defect in the change, a flaky test, an environment or dependency problem, or a broken pipeline configuration. Give the evidence for your answer. Then fix the smallest cause. Do not delete a test, skip it or mark it as expected to fail in order to make the build green, do not loosen a version constraint without saying what it was protecting, and do not touch unrelated jobs. Run the same command again and show it passing. Finish with the cause in two sentences, the diff, the commands you ran, and one change that would stop this class of failure returning. My first request is: "[paste the failing log here] The branch is [branch name]."
Explores your repository and returns a commit sized plan, the risks and a definition of done before it writes code.
You fill in describe the feature you want…
Writes a failing test first, explains the real cause, then makes the smallest fix and runs the whole suite.
You fill in describe the bug, how to…
Refactors behind passing tests, writing characterisation tests first and stopping the moment one of them fails.
You fill in file or modulethe reason
Reviews a branch or pull request for defects, risks and missing tests, with a verdict and no invented findings.
You fill in branch name, pull request…
One email a week: the best new prompts and one short guide. Unsubscribe any time. Privacy policy