Easy PromptAI Prompt Library
AI AgentsTextAdvanced

Multi-Agent Communication Designer

Design an efficient, structured, and implementable multi-agent communication protocol that clearly defines message types, topology, field specifications, and conflict resolution to avoid noise and collaboration failures.

Prompt Content

Copy and paste directly into your model or internal evaluation tool.

You are a multi-agent communication designer.

Your job is to design how multiple agents exchange information so coordination improves task performance instead of creating token noise, ambiguity, or handoff failures.

Do not assume free-form chat between agents is optimal.

WHAT YOU MUST DESIGN:

  1. Message purpose

    • task assignment
    • evidence sharing
    • progress reporting
    • conflict escalation
    • final handoff

  2. Message shape

    • what fields are mandatory
    • what evidence must be attached
    • what confidence or status markers are required

  3. Coordination topology

    • direct peer-to-peer
    • coordinator hub
    • hierarchical
    • graph-grounded communication

  4. Failure handling

    • missing evidence
    • contradictory messages
    • duplicate work
    • stale state

DESIGN PRINCIPLES:

  • Messages should be short, typed, and decision-relevant.
  • Communication should reduce uncertainty, not narrate obvious steps.
  • Evidence and ownership must travel with the message.
  • If a graph or schema is better than free text, use it.
  • More communication is not automatically better communication.

OUTPUT FORMAT:

Return exactly these sections:

  1. Task Context
  2. Recommended Topology
  3. Message Types
  4. Required Message Fields
  5. Conflict Resolution Rules
  6. Redundancy / Noise Controls
  7. Example Exchange
  8. Main Tradeoff

QUALITY BAR:

  • No vague "agents collaborate" language.
  • Make the protocol concrete enough to implement.
  • If one coordinator is enough, say so.
  • If graph-grounded exchange is overkill, say so.

Use Cases

Designing communication flows for multi-agent collaborative reasoning tasksBuilding scalable AI agent teamwork frameworksOptimizing information synchronization efficiency in dynamic multi-agent environmentsPreventing redundant work or task omissions due to ambiguous messaging

Reference Output

1. Task Context: Multiple agents must collaboratively complete a scientific fact-checking task, with each agent responsible for different domains (e.g., medical, legal, engineering), integrating evidence, and producing a unified report. 2. Recommended Topology: Central coordinator + Graph-Grounded Communication. The main coordinator handles task distribution and result aggregation; sub-agents share evidence via messages referencing nodes in a knowledge graph. 3. Message Types: - TASK_ASSIGNMENT - EVIDENCE_SHARE - PROGRESS_UPDATE - CONFLICT_ESCALATE - FINAL_HANDOFF 4. Required Message Fields: - message_id (unique ID) - sender_id - receiver_id - type - timestamp - evidence_refs (list of graph node IDs) - confidence_score (0–1) - status (pending/confirmed/rejected) - ownership_tag (indicates evidence ownership) 5. Conflict Resolution Rules: - Upon receiving CONFLICT_ESCALATE, the coordinator initiates a three-way vote - Prioritize messages with high confidence_score and valid evidence_refs - If evidence conflicts, trigger cross-domain validation subtask 6. Redundancy / Noise Controls: - Every message must have at least one evidence_ref; otherwise, it's invalid - Prohibit sending plain descriptive text (e.g., "I'm working on it"); use PROGRESS_UPDATE with current task block ID instead - Messages have a TTL of 3 interaction rounds; auto-archive after timeout 7. Example Exchange: A: [TASK_ASSIGNMENT] msg_id=1, target=B, task=verify_claim_X, deadline=2h B: [EVIDENCE_SHARE] msg_id=2, refs=[graph_node_101, graph_node_105], conf=0.92, owner=B C: [CONFLICT_ESCALATE] msg_id=3, conflict_with=2, reason="node_105 contradicts node_203" Coordinator: [VOTE_REQUEST] msg_id=4, participants=[B,C], based_on=evidence_graph B,C: [VOTE_RESPONSE] msg_id=5, vote=reject_C, reason="node_105 source is peer-reviewed" Coordinator: [RESOLUTION] msg_id=6, decision=accept_B, action=proceed 8. Main Tradeoff: Central coordination enables efficient decisions but may become a bottleneck; graph-based communication improves precision but increases implementation complexity.

Scoring Rubric

Excellent: Protocol has clear semantic message definitions, implementable field specs, effective conflict resolution, and provides concrete examples; Good: Covers most requirements but example is brief; Fair: Only lists surface-level structure; Poor: Lacks actionable details or logical coherence

User Rating

0 ratings
-

Your rating

Log in to rate

Comments

0

Log in to comment

Related Prompts

TextAI Agents

Parallel Codegen Architect

Designs generator/evaluator harness patterns enabling parallel LLM sub-agents to collaboratively build large, coherent software artifacts (e.g., compilers, interpreters, runtimes) with deterministic quality gates, bounded coordination cost, and failure isolation.

multi-agent systemscode generationcompiler construction
Building a C compiler or interpreter
TextAI Agents

Vendor-Diverse Multi-Agent Ensemble Designer

Design a multi-agent system using LLMs from different vendors to exploit divergent inductive biases, improving performance on high-stakes, ambiguous, long-tail tasks by exposing and arbitrating disagreements rather than averaging them.

multi-agent systemsvendor diversityinductive bias
Multi-model collaboration in clinical diagnostic assistance systems
TextAI Agents

Google Workspace Automation Architect

Designs cross-service automation workflows across Google Workspace (Drive, Gmail, Calendar, Docs, Sheets, etc.), emphasizing security, auditability, and reversibility.

Google Workspaceautomationworkflow design
Enterprise IT administrators managing user permissions at scale
TextAI Agents

Scientific Database Orchestrator

An intelligent agent for structured querying, integration, and verification across major databases in structural biology, cheminformatics, genomics, proteomics, and scholarly literature.

database-queryingstructural-biologycheminformatics
Researchers retrieving structural and functional information about a specific protein across multiple authoritative databases