Watch CI to green
Poll GitHub Actions until the latest run on your branch passes or fails.
/loop 3m check the latest GitHub Actions run on my current branch. Ping me on success or failure with a 1-line summary.
Production-ready templates. Copy, paste, swap the details โ you have a working loop.
Poll GitHub Actions until the latest run on your branch passes or fails.
/loop 3m check the latest GitHub Actions run on my current branch. Ping me on success or failure with a 1-line summary.
Dynamic-mode loop that polls a deploy queue and reports each stage.
/loop watch the staging deploy. Tell me when each stage starts and finishes. End the loop once production is live.
Run a known-flaky test until you have enough signal to trust the result.
/loop 90s run `npm test -- billing.spec.ts`. Stop after 5 consecutive passes or 3 failures.
Watch your PR list for new review comments, requests, and merge-conflicts.
/loop 15m list my open PRs. Report new review comments, requested changes, or merge conflicts since the last check.
Run a security review on every new PR to a sensitive folder.
/loop 30m for any new PR that touches src/auth/, run /security-review and post a comment if anything looks risky.
Nudge reviewers on PRs that have been open more than a day.
/loop 4h find PRs older than 24h with no review. Post a single polite reminder comment per PR.
Tail an error log and surface only patterns you have not seen before.
/loop 5m tail the last 500 lines of logs/app.log. Group new ERROR lines by stack trace. Report only fingerprints I have not seen before.
Fetch a dashboard number and alert on big swings.
/loop 10m fetch current p95 latency from Grafana. Alert if it jumps more than 30% vs. the previous reading.
Restart context: monitor a local dev server and report crashes.
/loop watch the dev server process. Capture stderr and tell me what crashed it if it dies.
Classify new threads, draft replies, label urgent ones.
/loop 30m read new Gmail threads from the last 30 minutes. Label by topic. Draft replies for the urgent ones.
Watch a single thread for new replies and summarize when you are away.
/loop 10m check the #incident-payments thread. Summarize anything new since the last check.
Look ahead at tomorrow's calendar and flag conflicts every evening.
/loop 1h between 5pm and 7pm, scan tomorrow's calendar. Flag overlaps and missing agendas.
Pair with /schedule for a once-a-day digest delivered every morning.
/schedule 8:30am summarize overnight: new PRs, failed CI runs, urgent emails, calendar for the day.
Gentle nudges if you have wandered off-task for too long.
/loop 25m peek at my active windows. If I am off-task for more than 10 minutes, surface a short reminder of my stated goal.
One small concept summarized each morning.
/schedule 7am summarize one concept from my reading list. 200 words max. Keep it concrete.
Notice new files in a shared directory and classify them.
/loop 5m list new files in ~/Drop. Move them to the right project folder based on filename.
Re-pull a remote dataset, update the local CSV, recompute summaries.
/loop 1h refresh sales.csv from the API. Update totals in summary.xlsx. Report diffs vs. yesterday.
Compare new screenshots in a folder against a reference and flag regressions.
/loop 10m compare new screenshots in screenshots/ to references/. Flag visual diffs > 5%.
Let Claude decide when to wake up โ best for unpredictable external state.
/loop poll the migration job. Wake fast while it is moving, sleep long while it is stable. End when it is done.
Use /schedule for unattended cloud-side automation.
/schedule every weekday at 9am: review yesterday's deploys and write a daily reliability note.
Wrap multiple commands into one loop.
/loop 30m run /verify, then /code-review on any unreviewed diff. Stop if both come back clean twice in a row.
Send us your favorite real-world Claude Loop and we will feature it in this gallery.
Submit a Loop โ