Agent instructions
AuraScore 81/100

Hierarchical Agent State Handoff and Memory Loss Analysis

Examine multi-agent context degradation, memory persistence failures, and state corruption across sequential and hierarchical delegation chains.

Use this template when multi-agent workflows suffer from token bloat, context loss during agent-to-agent delegation, or circular handoff loops. It provides a comprehensive state synchronization and context distillation assessment.

Template

Role: Lead Multi-Agent Systems Engineer specializing in distributed agent coordination and state orchestration.

Context

  • Multi-Agent Topology: {{orchestration_topology}}
  • Agent Role Manifest: {{agent_role_manifest}}
  • Shared Memory Architecture: {{shared_memory_protocol}}
  • Inter-Agent Handoff Schema: {{handoff_payload_schema}}
  • Execution Trace Failures: {{failure_scenario_traces}}
  • Data Retention Mandate: {{compliance_retention_rules}}

Task

Deliver an end-to-end multi-agent state handoff analysis that uncovers memory decay, context dilution, circular delegation, and payload desynchronization across agent transitions, establishing an optimized state retention model.

Method

  1. Analyze {{orchestration_topology}} to map critical path handoffs between parent, child, and peer agents.
  2. Trace context payload degradation across transitions using examples from {{failure_scenario_traces}}.
  3. Audit {{handoff_payload_schema}} for redundant token consumption, missing lineage metadata, and implicit assumptions.
  4. Evaluate {{shared_memory_protocol}} to detect race conditions, write conflicts, and stale state injection.
  5. Identify circular delegation patterns where sub-agents bounce unresolved subtasks back to orchestrators.
  6. Review compliance and redacting mechanisms against {{compliance_retention_rules}} across intermediate state saves.
  7. Design a minimal, deterministic context distillation contract for every transition edge in {{agent_role_manifest}}.
  8. Model token budget savings achieved by shifting from full-history concatenation to structured state diffs.

Constraints

  • MUST analyze every handoff edge specified in {{orchestration_topology}}.
  • MUST NOT recommend unconstrained conversational history passing between agents.
  • State contracts MUST remain compliant with {{compliance_retention_rules}}.
  • Provide concrete schema adjustments for {{handoff_payload_schema}}.

Output format

  1. Topology Context Flow Diagnostic (step-by-step audit of data fidelity loss per handoff)
  2. Memory Architecture Bottlenecks (assessment of {{shared_memory_protocol}} concurrency and persistence)
  3. Circular Routing & Deadlock Vulnerabilities (root cause analysis of unhandled edge cases)
  4. Optimized State Handoff Protocol (revised handoff schema with metadata and distillation logic)
  5. Instruction Directives for Delegation (system prompt instructions governing context summarization)

Self-review

  • Did the analysis pinpoint the exact mechanism causing context loss in {{failure_scenario_traces}}?
  • Are all handoff modifications compliant with {{compliance_retention_rules}}?
  • Does the proposed state contract eliminate token bloat while retaining execution-critical metadata?
AuraScore breakdown
81/100Provisional
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 engineering10/12 · Adequate

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.

ai-agents
agents-instructions
autonomous-agents-workflows
multi-agent
orchestration
context-engineering