Cookbook

Real apps do not stop at “define a task and call work.” They need jobs, databases, auth, rate limits, and traces that make sense at 2 a.m.

These recipes are copy-paste starting points for wiring CMDx into a Rails app. Each page stands on its own — open the one closest to what you are building today.

How each recipe is laid out

Most pages follow the same rhythm so you can skim:

1. Intro — what problem this solves and when to reach for it
2. Setup — middleware, initializer, or helper you install once
3. Usage — register it on a task and run an example
4. What to expect — sample SQL, JSON, or Result shapes (when it helps)
5. Notes — gotchas and “pick this vs that” tips

Rails and Active Record

• Active Record Database Transaction — commit every write together, or roll back as a unit
• Active Record Query Tagging — stamp SQL with the task name and trace ids
• Paper Trail Whatdunnit — tell Paper Trail which task changed a record

API and integration

• GraphQL Resolvers — keep mutations thin; let tasks do the work
• OpenAPI Schema Generation — generate request schemas from task inputs
• PubSub Task Chaining — trigger follow-on work with domain events

Cross-cutting concerns

• Background Execution — run tasks on a queue with Active Job or Sidekiq
• Store Wrappers — test doubles and instrumented store wrappers
• Flipper Feature Flags — gate tasks behind feature flags
• Pundit Authorization — authorize before work runs
• Timeout Guard — cap how long a task may run
• Sentry Error Tracking — send failures to Sentry with context
• OpenTelemetry Tracing — one span per task, nested like your chain