Web
AuraScore 83/100

Multi-Step Web Agent Execution Chain Resiliency Review

Analyze state degradation, cycle risks, and fallback reliability across multi-step autonomous web browser workflows.

Deploy this template when automated agent workflows experience state loss, endless navigation loops, or unhandled DOM drift. It produces a detailed reliability analysis covering DAG execution and state transition mechanics.

Template

Role: Staff Autonomous Systems Reliability Engineer with deep expertise in web automation workflow graphs and multi-step agent choreography.

Context

  • Workflow orchestration engine: {{orchestration_engine}}
  • Execution graph definition: {{workflow_dag_definition}}
  • Browser action primitive set: {{browser_action_primitives}}
  • Maximum session duration: {{session_timeout_ms}}
  • Current state recovery strategy: {{state_recovery_strategy}}
  • Execution trace telemetry: {{telemetry_source}}

Task

Conduct a rigorous reliability analysis of the multi-step agent workflow graph to pinpoint failure-prone state transitions, infinite loop vulnerabilities, DOM mutation edge cases, and deterministic recovery weaknesses.

Method

  1. Ingest the {{workflow_dag_definition}} to map every deterministic branch, probabilistic transition, and terminal state within {{orchestration_engine}}.
  2. Analyze trace telemetry from {{telemetry_source}} to calculate mean time between state desynchronizations and common failure nodes.
  3. Evaluate the robustness of {{browser_action_primitives}} against dynamic DOM mutations, lazy loading, and dynamic hydration delays.
  4. Audit cycle-detection safeguards to determine how circular execution paths are caught before breaching {{session_timeout_ms}}.
  5. Stress-test the {{state_recovery_strategy}} against unhandled modal dialogs, network degradation, and rate-limiting responses.
  6. Quantify context window bloat across iterative chain steps to identify state compression bottlenecks.
  7. Formulate a resilient graph topology with idempotent fallback hooks, self-healing retries, and strict progress invariants.

Constraints

  • MUST calculate empirical failure probabilities for each node transition in {{workflow_dag_definition}}.
  • MUST NOT assume deterministic DOM state across consecutive web navigation steps.
  • Recommendations must be implementable within the architectural constraints of {{orchestration_engine}}.
  • Analysis must define deterministic circuit-breaker criteria for every looping primitive.
  • Keep recommendations focused on reliability, idempotency, and session integrity.

Output format

  • Executive Summary (under 200 words)
  • Workflow Node Criticality & Vulnerability Table (Node, Failure Mode, Impact, Severity)
  • Cycle and Timeout Vulnerability Analysis (detailed step breakdown)
  • Resiliency & Recovery Architecture (concrete graph revisions and state persistence rules)
  • Verification Plan (synthetic test scenarios and assertion criteria)

Self-review

  • Verify that each failure mode references specific primitives in {{browser_action_primitives}}.
  • Confirm that cycle analysis accounts for the constraint imposed by {{session_timeout_ms}}.
  • Ensure recovery strategies do not introduce non-idempotent duplicate side effects.
AuraScore breakdown
83/100Provisional
Instruction clarity15/15 · Strong

Explicit role, a named task, and discrete steps the model can follow.

Context architecture12/12 · Strong

Background, inputs and variables the model needs before it starts.

Constraint engineering12/12 · Strong

Hard boundaries — what the model must and must not do.

Output specification6/14 · Thin

A named, field-level shape for the response.

Reasoning structure10/10 · Strong

Ordered work items that force analysis before an answer.

Model compatibility10/10 · Strong

Length and structure that travel across frontier models.

Token efficiency5/10 · Thin

Signal density — instruction weight without padding.

Reusability7/7 · Strong

Documented variables so the scaffold adapts to new inputs.

Robustness5/5 · Strong

Quality bar, assumptions and behaviour when inputs are thin.

Observed performance1/5 · Thin

How much real usage the template has behind it.

developers
developers-web
autonomous-agents-workflows
workflow-chains
browser-automation
agentic-ai