Uses your own Claude Code subscription — no token markup

Cron for Claude Code.

Schedule claude -p jobs that run on your machine, on your Claude plan. We handle the cron, retries, history, and notifications — your subscription, your files, your secrets never leave your laptop.

Start free Read the docs

Free forever for hobby use · No credit card · Cancel anytime

claude-scheduler · dashboard
Schedules
17
Online runners
2/2
Jobs · 24h
284
+12%
Recent jobs
iterateFindEat2 min agosucceeded
iterateLawHero2 min agosucceeded
ideate-saasansicode2 min agorunning
iterateatvref2 min agoqueued
Why claude-scheduler

Orchestrate the Claude CLI
you already pay for.

Use your Max / Team / API plan as-is. We never see your tokens, never charge per-job markup, never leave your prompts on someone else's GPU.

Use your existing Claude plan

The runner calls claude -p on your machine using whichever Claude Code login is already there. No second subscription, no token markup, no per-call API key juggling. If your plan covers it, it covers our jobs.

Your machine, your files

Prompts run where your code already lives — your laptop, a Mac mini, a Linux box. Working directories stay local; secrets never touch our infra; output streams back to the dashboard.

Production-grade scheduling

Cron + one-shot triggers. Skip-if-running, retries with backoff, lease-based recovery on crash, optional Claude --resume to keep the conversation warm across runs.

Cron you can trust

Per-schedule timezone, validated expressions, live next-fire preview, complete run history. Logs, tokens, cost — everything from every run, kept around for as long as your plan allows.

Notifications that don't spam

Telegram alerts on never / on-failure / on-success / always. Decide per-schedule, override per-project. Quiet when you want, loud when it matters.

Built for teams

Multi-org with roles. API keys for CI. Audit log on every mutation. Sign in with email/password, Google, or GitHub.

How it works

Three steps. Done.

  1. 1

    Add a runner

    From the dashboard's Runners page, click Create runner. You'll get a one-time install command pre-filled with your token — paste it on any Mac or Linux box that already has the Claude CLI logged in.

  2. 2

    Write a schedule

    Pick a project (or skip), a cron expression (we preview the next 5 fires), and the prompt. Working dir, runner tags, retry policy, Telegram notify — all optional, all overridable.

  3. 3

    Watch it run

    Live stream-json activity timeline, stdout / stderr, exit code, token usage, cost per run. Drill into any run from the Jobs tab. Replay anything that failed.

Full walkthrough → Read the docs
Vs. running cron yourself

crontab + ssh works. Until it doesn't.

crontab on your Mac
  • ×Job fires while laptop is asleep — silently lost.
  • ×No history. Did Tuesday's run succeed? Cat /var/log/…
  • ×One mismatched flag stalls forever waiting for permission.
  • ×You SSH into 3 machines to check what ran where.
  • ×No retries. No backoff. No notifications.
  • ×To pause one job you edit a textfile in vi.
claude-scheduler
  • Server-side cron. Fires whether or not the laptop is awake — runner picks up when next online.
  • Every run kept: prompt, stdout, stderr, tokens, cost. Filter by status, project, schedule.
  • --dangerously-skip-permissions on by default for unattended runs.
  • One dashboard across every machine. Click any job to see what happened.
  • Configurable retries with backoff. Telegram pings when something breaks.
  • Toggle a schedule on/off in one click. Edit the prompt, change the cron, no SSH.

Free to start. Honest pricing.

Free tier covers hobby use. Paid plans scale schedules, runners, and history — nothing else.

See plans Start free
  • No credit card
  • Cancel anytime
  • Your runner · your keys
  • Stop billing in 1 click