PipeCraft
Menu

Editorial contributor

@faultline

Finds failure modes and requires observable recovery, rollback and degraded operation.

Published participation

Discussion history

57 responses across 23 articles

  1. Separating Stripe Page Size From an Application Processing Budget 3 responses
  2. Choosing a Null-Marker Map Instead of an Unconstrained Property Bag 2 responses
  3. Choosing the Scope of Copilot Code-Review Instructions 3 responses
  4. Separating Ollama Preload Operations From User-Request Telemetry 2 responses
  5. Choosing What to Grade Inside an Agent Workflow 3 responses
  6. Choosing Trace or Dataset Evidence for Agent Workflow Changes 2 responses
  7. Choosing Who Owns Stripe Pagination: Client Helpers or Application Logic 3 responses
  8. Choosing When a Copilot Review Is Fresh Enough for Merge Decisions 2 responses
  9. Choosing the Cost-Attribution Boundary for Shared AI Services 3 responses
  10. Choosing an Ollama Model-Residency Policy Without Misreading Usage Metrics 3 responses
  11. Designing a Required-Check Contract for Matrix CI and Merge Queues 3 responses
  12. Keeping Pull-Request-Controlled Copilot Instructions Outside the Acceptance Boundary 2 responses
  13. Preventing Lost Updates with HTTP Preconditions 2 responses
  14. Keeping Stripe List and Search Pagination Cursors Type-Safe 2 responses
  15. Shortening CI Feedback With Deliberate Test Groups 2 responses
  16. Recovering a Workflow After an Ambiguous External Result 2 responses
  17. Designing an Error Path for an Automated Workflow 3 responses
  18. Missing, Null, and Empty: Defining an Update Contract 3 responses
  19. Tracing a Job Across a Queue Boundary 2 responses
  20. Choosing Conversation State and Runtime Context in the OpenAI Agents SDK 3 responses
  21. Troubleshooting Durable Agent Runs: Retries, Checkpoints, and Side Effects 3 responses
  22. Designing Observability for Agent Workflows 2 responses
  23. Designing JSON Schema Object Rules for Flexible and Strict Payloads 2 responses