More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+22 more

    Loudcron: Scheduled jobs that cannot fail in silence

    1

    Installs a recurring job with preflight checks, a lock, retries and alerts on failure, ships a pruner in the same change, then confirms real output from a manual run and the first two scheduled runs.

    $12

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Loudcron: Scheduled jobs that cannot fail in silence

    Loudcron: Scheduled jobs that cannot fail in silence

    Example session with this skill installed

    Loudcron: set up a nightly refresh for our product feed on Vercel cron. It pulls from 3 supplier APIs and writes to Postgres.

    • Read your context and instructions
    • Compiled the loudcron

    Installed refresh-feed as one unit. Runner: Vercel cron, because it must run when your laptop is off.

    Job

    • scripts/refresh-feed.mjs: idempotent, --dry-run, exits non-zero if any of the 3 suppliers fails
    • Route /api/cron/refresh-feed, 02:00 daily, max duration 60s
    • Auth fail-closed: rejects when CRON_SECRET is empty or mismatched; route added to the auth test matrix

    Preflight: CRON_SECRET, 3 supplier API keys and DB URL present; fails on the first missing one.

    Guards: lock row so manual and scheduled runs never overlap · 3 attempts per supplier · 45s timeout.

    Alert: Telegram bot message on UP to DOWN and DOWN to UP, plus when feed data is older than 26h.

    Pruner (same change): feed_runs table kept 30 days by a pg_cron job; per-run log rows rotated by size.

    Registry

    LabelRunnerScheduleStatusRe-enable
    refresh-feedVercel cron02:00 dailyactiverestore entry in vercel.json

    Verify

    RunExitSupplier countsRows written
    manual0412 / 388 / 97897
    scheduled 10415 / 388 / 99902
    scheduled 21416 / 0 / 99515

    Run 2 counted as failed (supplier B skipped) and the alert fired. Cause: supplier B key expired.

    Next steps

    1. Renew the supplier B key and trigger a manual run.
    2. Watch the next scheduled run for 3 non-zero counts.
    3. Retire the old local crontab entry that still writes to prod.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    A scheduled job can look healthy for weeks while doing nothing. A secret was never set, a login session expired, or one of four sources quietly stopped, and the only signal is a log file nobody reads, growing without limit. Loudcron installs a recurring job as one unit (runner, schedule, guards, alert, pruner and retire path) and gives a safe ritual for sweeping what piles up.

    What it does

    • Job script rules: idempotent, --dry-run, non-zero exit on full or partial failure.
    • Runner choice: local (launchd, cron, systemd timer) only when it needs your machine; cloud (Vercel cron, GitHub Actions, a scheduled cloud machine, pg_cron) when it must be reliable.
    • Preflight checks secrets, session, account and permissions and fails loudly on the first gap.
    • Guards: lock file, max attempts, timeout, stale-PID detection and an end date for batch jobs.
    • Alerts on state changes through a desktop notification and a Telegram bot.
    • Pruner in the same change: log rotation, table retention, or move-only with a manifest.
    • Optional audit trail: a hash-chained JSONL log for jobs that act on their own, with rotation and no secret values.

    How it works

    1. Install: write the script, pick and register the runner, with fail-closed auth for cloud cron.
    2. Harden: preflight, guards, alert and pruner, plus a registry entry with re-enable steps.
    3. Verify: a manual trigger, then real output from the first two scheduled runs.
    4. Diagnose: log and exit code first, then the runner and system checks.
    5. Sweep and retire: read-only audit, keep, stop or remove by bucket, move-only, manifest.

    Frameworks & tools

    launchd, cron, systemd timers, Vercel cron, GitHub Actions, supercronic, pg_cron, osascript and the Telegram Bot API.

    Why this beats prompting it yourself

    A quick prompt writes a crontab line. This skill does not count a job as done until an alert fired in a test, a pruner shipped with it, and real output landed. Its guards come from real runs, such as a cloud cron that failed 178 times because a secret was never set.

    Use cases

    • Data refresh or an hourly scrape with per-source counts.
    • Deferred publishing from a queue.
    • Ops watchdog for uptime, freshness and budget.
    • Monthly tidy of files and cloud clutter.

    Known limitations

    Not for one-off tasks. Local jobs run only while the machine is on. Stopping processes, retiring jobs or emptying trash happens per bucket after you approve.

    How to use

    Skills on Agensi run inside your AI agent. Copy the prompt, paste it into your agent, and it connects to Agensi and installs this skill for you. If a payment is needed, your agent shows it to you first.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill passes our trust checks below.

    Security scanned

    Passed our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scanned

    Verified clean 1 day ago

    • Passed all security checks, Safe to install

    Requires

    launchd
    cron
    systemd timer
    Vercel cron
    GitHub Actions
    pg_cron
    Telegram Bot API
    macOS

    Listed1 day ago

    What's inside

    Frequently Asked Questions