Welcome to claude-scheduler

claude-scheduler runs claude -p jobs on machines you control at times you pick. We handle the cron, retries, logs, and notifications. Your Claude subscription, your machine, your files — we never touch any of those.

The three concepts

  • Runner — a small Node daemon you install on a machine where the Claude CLI is already logged in. It polls our API, claims due jobs, runs claude -p in the directory you chose, and streams the result back.
  • Project — a named bundle of defaults: working directory, runner tags, Claude args, notify policy, effort level. Use it when many schedules share the same setup so you don't repeat yourself.
  • Schedule — a cron expression (or one-shot time) + prompt. Inherits from its project; can override any field. Each fire becomes a Job, queued on our side and claimed by a runner.

Quickstart (5 minutes)

  1. Sign up — we create an organization for you. No credit card.
  2. Go to Runners → Create runner → name it (my-mac is fine). You'll get a one-time token starting with csr_live_.
  3. Install the runner on the machine you want jobs to run on. Reload Runners — your runner shows up as online within a few seconds.
  4. (Optional) Make a Project if you want schedules to share defaults. Otherwise schedules can be ad-hoc.
  5. Create your first schedule — pick a cron expression (we preview the next 5 fires), write the prompt, save. Wait for the next fire or click Run now.

Architecture

┌────────────────────┐                ┌──────────────────────┐
│  Web UI            │   HTTPS/JSON   │  Postgres            │
│  Next.js · Stripe  │◀──────────────▶│  schedules, jobs,    │
│  · auth · cron     │                │  runners, audit, …   │
└──────────┬─────────┘                └──────────────────────┘
           │ HTTPS poll
           ▼
   ┌────────────────┐ ┌────────────────┐
   │  runner #1     │ │  runner #2     │  …on your machines
   │  claude -p $@  │ │  claude -p $@  │
   └────────────────┘ └────────────────┘

Where to go next

  • Install a runner — daemonising, launchd / systemd, multiple runners on one host.
  • Schedules — cron syntax, timezones, retries, skip-if-running, Claude session continuation.
  • Projects — when to use them, how inheritance works, override patterns.
  • Notifications — Telegram setup, per- schedule policies.
  • REST API — orchestrate from CI / scripts with a csk_live_ API key.