Build or review dygo durable Jobs, Job Executions, queues, workers, retries, idempotency, and Schedules. Use for background or recurring business work.
Installs into .claude/skills of the current project.
Are you the author of Dygo Jobs And Schedules?
Add the live security badge to your README. It updates with every re-scan.
[](https://www.skillsdirectory.com/skills/hapyco-dygo-jobs-and-schedules)
---
name: dygo-jobs-and-schedules
description: Build or review dygo durable Jobs, Job Executions, queues, workers, retries, idempotency, and Schedules. Use for background or recurring business work.
---
# dygo Jobs And Schedules
Use Jobs for durable work outside a user request. Use Schedules to create recurring Job Executions.
## Start
Read the relevant sections of `docs/jobs.md`, `docs/schedule.md`, `docs/queues.md`, and `docs/sdk.md`. Inspect the target App's Jobs and queue configuration.
## Rules
- Identify a Job by `<app>/<job>`. Do not use its label as identity.
- Keep the Job payload explicit, stable, and safe to persist.
- Use an idempotency key when duplicate execution can harm the business. Define how retries recover partial completion; a deduplication key alone does not ensure completion.
- Job handlers run outside request transactions. A Record transaction does not automatically bind the Job or Notification service to that transaction. Trace the supplied services before relying on atomic writes.
- For a Record plus queued-work operation, check failure between persistence and enqueueing. A retry must complete the missing work or observe a complete prior operation.
- For transaction or retry changes, use a focused failure-path check that can detect partial commits. A success-only fake does not prove rollback or concurrency behavior.
- Make retry behavior match the failure type. Do not retry permanent validation failures.
- Keep handlers observable through result data, errors, timing, and persisted Logs.
- Keep queue and concurrency choices explicit when ordering or resource limits matter.
- Remember that `dygo serve` does not process Jobs. Production needs a separate worker process.
- Treat file-defined, Studio-defined, and system-owned Schedules according to the documented source contract.
- Do not add a second background execution mechanism inside an App.
## Check
Validate registration and runner wiring. Use focused execution inspection. Confirm worker behavior when changes affect claims, retries, Schedules, or concurrency.