Job orchestration

Your cron jobs grew up. Give them a graph.

Spindle replaces scattered crontabs and one-off CI jobs with a single scheduler: jobs declare what they depend on, Spindle decides what runs when, retries what failed, and tells you before a late run becomes an incident.

• Schedules in plain cron syntax or as an interval — plus event and manual triggers.
• Dependencies between steps, so a report never reads a half-finished table.
• Retries with backoff per step, not blunt retries of the whole pipeline.
• Agents run in your own network; Spindle never needs inbound access to your boxes.

Read the quickstart   See what's included →

A workflow is a file

Workflows live in your repository next to the code they run. Spindle picks up the definition from a Git ref or from the CLI, so a change to a schedule goes through the same review as any other change.

# spindle.yaml
workflow: warehouse-nightly
schedule: "20 2 * * *"      # 02:20, agent timezone
timezone: Europe/Amsterdam
late_after: 45m              # alert if not finished by then

steps:
  - name: load-orders
    run: "./bin/extract orders"
    retries: { max: 3, backoff: 30s }

  - name: load-refunds
    run: "./bin/extract refunds"
    retries: { max: 3, backoff: 30s }

  - name: build-export
    needs: [load-orders, load-refunds]
    run: "./bin/export --day=$SPINDLE_RUN_DATE"
    timeout: 20m

Everything else — history, logs, the dependency graph, the alert when build-export is still running at 03:05 — comes from that file without extra configuration.

Recent runs