Loops

Manually started loop that fixes and re-audits until changed UI has no serious accessibility violations.

Manually started loop that fixes errors and rebuilds until the production build succeeds.

A manually started loop that adds tests until coverage reaches the target and all tests pass.

Polls the latest CI run on the current branch until its conclusion is success.

Repeats diagnose, fix and push until the latest CI run on the branch succeeds.

Interval-driven check of PRs labeled codex-watch, keeping CI green and branches current, escalating when needed.

Runs the formatter again and again until it leaves no file changes behind.

Works through the stories in .ralph/prd.json one round at a time, gated by tests, lint and build.

Separates flaky tests from real regressions and fixes only the real ones.

Blocks git commits while tests fail and lets them through once tests pass.

Reviews recent changes for unneeded or off-convention code, then confirms with lint and tests.

Manually started loop that gets tests and lint passing while avoiding failure patterns seen in earlier rounds.

Re-checks health endpoints after a deploy on an interval until every one responds successfully.

Manually triggered loop that brings docs back in line with code changes on a branch.

Audits npm dependencies, applies safe fixes and runs tests until no high or critical issues remain.

Builds a feature test-first: write a failing test, make it pass, then refactor with the suite green.

Periodically lists PRs labelled codex-watch, checks their status and reports on each one.

Edits openapi.yaml round by round until Redocly lint passes and the spec matches the routes.

Runs build, lint and test checks under independent verification until all of them pass.

Manual loop that reruns knip and cleans up findings until it exits 0, within 5 rounds.

Runs contract tests and fixes the implementation until the API matches its published contract.

Implement, test and update a PR round by round until every CI check passes.

After a ship, compare commits since the last tag and fill in CHANGELOG.md's [Unreleased] entries.

Reproduce a failing test, reflect on each failure, and patch until it passes.

Manually started loop that reruns Prisma migrations until the status check comes back clean.

Runs the smoke suite right after each merge or rebase to catch regressions early.

Reruns the staging smoke checks until the whole checklist passes.

Runs repeated self-review passes over the branch diff against main until no critical findings remain.

Fixes code round by round until lint and the type check both come back clean.

Runs tests related to each batch of edited files and blocks further work until they pass.

Audit npm dependencies on a schedule and post a weekly summary with a remediation plan.

Weekly interval routine that lists outdated npm packages and posts an upgrade-focused summary.

Upgrades critical outdated dependencies one at a time, verifying each with tests and a build.

Proves a bug's root cause round by round with a minimal repro script.

Repeats fix-and-retest rounds with npm test until the whole suite passes.

Implements spec.md requirements one by one, verified with npm test, until all are checked off.

Rebases a branch onto main, resolves conflicts round by round and confirms with tests.

Manually started loop that rebuilds and measures until the client bundle fits its size-limit budget.

Reruns Playwright visual tests until the UI matches the expected baselines.

Manually started loop that fixes and reruns end-to-end tests until the E2E command exits 0.