AGNT5 cookbooks
End-to-end production AI workflow patterns for durability, debugging, evals, and recovery.
| Cookbook | Topics | Updated |
|---|---|---|
| Debug and replay a failed AI workflow Inspect a failed run, replay it locally, patch the failure, and resume from the failed step. | May 13, 2026 | |
| Build a durable human-approval AI workflow Pause for approval, survive worker restarts, and execute the final side effect exactly once. | May 13, 2026 | |
| Retry AI workflow steps without duplicate side effects Use idempotency keys and journaled receipts so retries do not duplicate emails, tickets, or payments. | May 13, 2026 | |
| Turn a failed production AI run into an eval Capture a bad production run, convert it into an eval case, and compare fixed prompts before release. | May 13, 2026 | |
| Build a durable research agent with approval and recovery Checkpoint search, extraction, artifacts, and final human approval across a long-running report workflow. | May 13, 2026 | |
| Debug AI workflows with traces, not scattered logs Compare log-only debugging with workflow-native traces that preserve inputs, outputs, retries, and state. | May 13, 2026 | |
| Run any AI agent framework on a durable workflow runtime Wrap LangGraph, OpenAI Agents SDK, Vercel AI SDK, or plain Python in durable AGNT5 steps. | May 13, 2026 | |
| Build a customer support agent Search docs, draft replies, wait for review, and turn support outcomes into eval cases. | May 13, 2026 | |
| Build a deep research agent Run long research jobs with search tools, streamed progress, trace inspection, and completion notices. | May 13, 2026 | |
| Build a document processing pipeline Extract structured fields, validate them, pause for review, and retry failed document steps safely. | May 13, 2026 | |
| Build a data extraction workflow Call tools, force JSON outputs, recover from malformed responses, and inspect every extraction step. | May 13, 2026 | |
| Build a RAG chatbot with memory Retrieve knowledge, preserve user context, isolate tenants, and trace each answer back to evidence. | May 13, 2026 | |
| Build a webhook triage agent Receive events, deduplicate delivery attempts, run async triage, and expose a trace per webhook. | May 13, 2026 | |
| Build a model comparison workflow Run the same case through multiple models, score outputs, and promote the release candidate. | May 13, 2026 |