Skip to main content

Writing workflows

Husky workflows live in husky.yaml. Each job should answer four questions clearly:

  1. what does it do?
  2. when does it run?
  3. what does it depend on?
  4. what happens if it fails?

Minimal job​

jobs:
  hourly_ping:
    description: "Check endpoint reachability"
    frequency: hourly
    command: "./scripts/ping.sh"

Typical production-ready job​

jobs:
  weekday_backup:
    description: "Back up the primary database"
    frequency: on:[monday,tuesday,wednesday,thursday,friday]
    time: "0100"
    command: "./scripts/backup.sh"
    timeout: "30m"
    retries: 2
    retry_delay: exponential
    on_failure: alert
    tags: [critical, maintenance]
    env:
      DB_HOST: "${env:BACKUP_DB_HOST}"

Workflow authoring checklist​

When defining a job, consider:

  • use a specific, human-readable description
  • choose a readable frequency
  • add time when wall-clock scheduling is used
  • define timeout
  • define retries and retry_delay
  • choose on_failure
  • tag jobs for bulk operations
  • add notify when the job matters operationally
  • use healthcheck when exit code alone is not enough
  • use output when downstream jobs need produced values

Defaults​

The defaults: block keeps configuration concise and consistent.

Common defaults include:

  • timeout
  • retries
  • retry_delay
  • on_failure
  • default_run_time
  • timezone