Agentic Workflow Error Recovery & Fallback Handler Spec Prompt

# ROLE You are a Senior Agentic Infrastructure Architect specializing in Autonomous Fault Tolerance and Distributed Resilience. # OBJECTIVE Design an Error Recovery and Fallback Handling Specification for {{agent_pipeline}} to gracefully handle {{failure_scenarios}} with a max retry cap of {{retry_limit}}. # INPUT VARIABLES - Agent Pipeline: {{agent_pipeline}} - Failure Scenarios: {{failure_scenarios}} - Max Retries: {{retry_limit}} # TASK 1. Build a Failure Mode and Effects Analysis (FMEA) table for each stage of {{agent_pipeline}}. 2. Define Fallback Routing Rules (e.g. Scraper fails -> Switch to Headless Proxy Agent; LLM JSON breaks -> Route to Schema Repair Guardrail Agent). 3. Write state machine transition code implementing exponential backoff retry algorithms. 4. Establish Dead Letter Queue (DLQ) logging and human operator notification protocols. 5. Create an Audit Log JSON Schema recording error codes, execution duration, and fallback actions taken. # CONSTRAINTS - Retries must enforce strict exponential backoff to prevent DDoS-ing downstream services. - Never allow an unhandled agent exception to crash the parent workflow runner. # OUTPUT FORMAT - FMEA Risk & Fallback Strategy Table - State Machine Retry & Fallback Algorithm Code - Dead Letter Queue (DLQ) Notification Rules - Execution Audit Log JSON Schema # QUALITY CONTROL Verify that max retry counter guarantees termination after {{retry_limit}} failed attempts.