Skill: backend/job-worker-orchestration

Skill: backend/job-worker-orchestration

Purpose

Implement background work safely using the repo’s job system (queue/cron/worker) with:

• explicit idempotency guarantees
• bounded retries and backoff
• durable state and re-entrancy safety
• observability (logs/metrics/traces) sufficient to operate it

Use when work is async, scheduled, long-running, or must survive restarts.

Inputs

• Job intent (what it does, why it’s async)
• Trigger type:
  • event-driven (enqueue)
  • scheduled (cron)
  • periodic poller
• Payload shape (fields + types)
• Idempotency key strategy (preferred) or invariants that ensure idempotency
• Failure expectations:
  • transient vs permanent errors
  • acceptable delay
• Repo profile (preferred): <repo>/REPO_PROFILE.json

Outputs

• Job definition/handler
• Enqueue/schedule wiring
• Idempotency mechanism:
  • idempotency key + dedupe storage OR
  • safe “exactly-once-ish” strategy documented
• Retry/backoff policy (bounded)
• Dead-letter or quarantine path if repo supports it
• Tests:
  • unit tests for handler logic
  • integration tests for enqueue/execute when infrastructure exists
• Minimal operational notes (inline comments or docs per repo norms)