Retry & Checkpoint Directives

examples/16-retry-checkpoint.halo

How to add resilience with @retry and @checkpoint WHY: Handle transient failures, save intermediate state

Open in the playground →

// ============================================================
// STEP 16: Retry & Checkpoint Directives
// ============================================================
// LEARN: How to add resilience with @retry and @checkpoint
// WHY:  Handle transient failures, save intermediate state
// ============================================================

agent Resilient [openai/gpt-4o, temp: 0.7]

// Retry directive: retry up to 3 times, wait 5s between retries
@retry(3, 5s)
// Checkpoint: save state after this pipeline completes
@checkpoint
pipeline RobustCall(data: String) -> Result:
  data
    |> Resilient.prompt("Process: {{data}}")
    // If this fails, retries happen automatically
    |> System.Print()

// Retry syntax: @retry(count, delay)
//   count: number of retry attempts
//   delay: time to wait between retries (e.g., 5s, 1m)
//
// Checkpoint syntax: @checkpoint
//   Saves pipeline state for recovery
//   (Implementation: logs checkpoint, full persistence pending)