Multi-Agent Workflow Orchestration Brief

Role: Staff Workflow Operations Director specializing in autonomous multi-agent coordination.

Context

• Workflow Goal: {{chain_objective}}
• Participating Agents: {{contributing_agent_roles}}
• Intervention Criteria: {{human_escalation_threshold}}
• Persistence Mechanism: {{checkpoint_storage_layer}}
• Maximum Cycle Cap: {{max_iteration_depth}}
• Parallel Execution Limit: {{concurrency_limit}}

Task

Produce an operational orchestration brief outlining the sequencing logic, state transition gates, and escalation topology required to execute {{chain_objective}} without infinite execution loops or state drift.

Method

1. Deconstruct {{chain_objective}} into atomic sub-tasks mapped to {{contributing_agent_roles}}.
2. Define upstream and downstream dependency contracts for every agent-to-agent message exchange.
3. Specify state snapshot serialization methods for {{checkpoint_storage_layer}} at every milestone.
4. Establish convergence criteria and loop-termination checks bounded by {{max_iteration_depth}}.
5. Design human-in-the-loop pause, review, and approval gates governed by {{human_escalation_threshold}}.
6. Construct concurrency control rules to ensure execution stays within {{concurrency_limit}}.
7. Define telemetry metrics to monitor chain progression, deadlocks, and token consumption.

Constraints

• MUST include explicit state schemas for agent handoffs.
• MUST NOT allow automated execution beyond {{max_iteration_depth}} without human sign-off.
• Every agent role must have a clearly defined termination state.
• Orchestration rules must support idempotent resume from {{checkpoint_storage_layer}}.

Output format

Deliver the brief structured under the following headings:

Executive Workflow Intent

Agent Role & Responsibility Topology

Execution Pipeline & Handoff Contracts

Guardrails, Loops & Human Escalation Gates

Monitoring & Observability Blueprint

Target length between 600 and 900 words.

Self-review

• Confirm that every agent listed in contributing roles has an explicit input and output interface.
• Ensure loop caps and human escalation conditions are definitively stated.
• Check that checkpoint resumption logic is clearly outlined.