viktorbezdek Multi-Agent Architecture Patterns
Multi-agent architectures distribute work across multiple language model instances, each with its own context window. When designed well, this distribution enables capabilities beyond single-agent limits. When designed poorly, it introduces coordination overhead that negates benefits. The critical insight is that sub-agents exist primarily to isolate context, not to anthropomorphize role division.
When to Use
- Single-agent context limits constrain task complexity
- Tasks decompose naturally into parallel subtasks
- Different subtasks require different tool sets or system prompts
- Building systems that must handle multiple domains simultaneously
- Scaling agent capabilities beyond single-context limits
- Designing production agent systems with multiple specialized components
When NOT to Use
- Agent memory or persistence across sessions (use memory-systems)
- Tool design or tool interfaces (use tool-design)
- Hosted agent infrastructure or sandboxed VMs (use hosted-agents)
- BDI cognitive models or mental state modeling (use bdi-mental-states)
- Simple tasks that fit within a single context window (no need for multi-agent)
Decision Tree
Do you need multiple agents?
│
├─ Single-agent context limits reached?
│ ├─ YES → Continue
│ └─ NO → Single agent is simpler and cheaper; stay with it
│
├─ How should agents coordinate?
│ ├─ Central control needed? → Supervisor/Orchestrator
│ │ ├─ Need strict workflow control? → Supervisor with plan enforcement
│ │ └─ Need human-in-the-loop? → Supervisor with approval gates
│ ├─ Flexible exploration needed? → Peer-to-Peer/Swarm
│ │ ├─ Tasks have emergent requirements? → Swarm with handoff protocols
│ │ └─ Breadth-first search? → Swarm with convergence constraints
│ └─ Large-scale with layers? → Hierarchical
│ ├─ Strategic + planning + execution layers → 3-tier hierarchy
│ └─ Enterprise workflows with management levels → Match org structure
│
├─ How should context be isolated?
│ ├─ Sub-agent needs full understanding? → Full context delegation
│ ├─ Sub-task is well-defined? → Instruction passing only
│ └─ Need shared state without context bloat? → File-system memory
│
└─ Consensus mechanism needed?
├─ Quick decision? → Weighted voting by confidence
├─ High-stakes accuracy? → Debate protocol (adversarial critique)
└─ Detecting sycophancy? → Trigger-based intervention
Core Concepts
Multi-agent systems address single-agent context limitations through distribution. Three dominant patterns exist: supervisor/orchestrator for centralized control, peer-to-peer/swarm for flexible handoffs, and hierarchical for layered abstraction. The critical design principle is context isolation—sub-agents exist primarily to partition context rather than to simulate organizational roles.
Effective multi-agent systems require explicit coordination protocols, consensus mechanisms that avoid sycophancy, and careful attention to failure modes including bottlenecks, divergence, and error propagation.
Detailed Topics
Why Multi-Agent Architectures
The Context Bottleneck Single agents face inherent ceilings in reasoning capability, context management, and tool coordination. As tasks grow more complex, context windows fill with accumulated history, retrieved documents, and tool outputs. Performance degrades according to predictable patterns: the lost-in-middle effect, attention scarcity, and context poisoning.
Multi-agent architectures address these limitations by partitioning work across multiple context windows. Each agent operates in a clean context focused on its subtask. Results aggregate at a coordination layer without any single context bearing the full burden.
The Token Economics Reality Multi-agent systems consume significantly more tokens than single-agent approaches. Production data shows:
| Architecture | Token Multiplier | Use Case |
|---|---|---|
| Single agent chat | 1× baseline | Simple queries |
| Single agent with tools | ~4× baseline | Tool-using tasks |
| Multi-agent system | ~15× baseline | Complex research/coordination |
Evaluation work on browsing and research tasks repeatedly shows that token usage, tool-call strategy, and model choice explain much of the variance in agent performance. This validates the multi-agent approach of distributing work across agents with separate context windows to add capacity for parallel reasoning.
Critically, upgrading the model or reasoning configuration can outperform simply increasing token budgets. Treat model selection, reasoning mode, and multi-agent architecture as complementary strategies, and verify the trade-off on the actual task suite instead of relying on dated benchmark claims.
The Parallelization Argument Many tasks contain parallelizable subtasks that a single agent must execute sequentially. A research task might require searching multiple independent sources, analyzing different documents, or comparing competing approaches. A single agent processes these sequentially, accumulating context with each step.
Multi-agent architectures assign each subtask to a dedicated agent with a fresh context. All agents work simultaneously, then return results to a coordinator. The total real-world time approaches the duration of the longest subtask rather than the sum of all subtasks.
The Specialization Argument Different tasks benefit from different agent configurations: different system prompts, different tool sets, different context structures. A general-purpose agent must carry all possible configurations in context. Specialized agents carry only what they need.
Multi-agent architectures enable specialization without combinatorial explosion. The coordinator routes to specialized agents; each agent operates with lean context optimized for its domain.
Architectural Patterns
Pattern 1: Supervisor/Orchestrator The supervisor pattern places a central agent in control, delegating to specialists and synthesizing results. The supervisor maintains global state and trajectory, decomposes user objectives into subtasks, and routes to appropriate workers.
User Query -> Supervisor -> [Specialist, Specialist, Specialist] -> Aggregation -> Final Output
When to use: Complex tasks with clear decomposition, tasks requiring coordination across domains, tasks where human oversight is important.
Advantages: Strict control over workflow, easier to implement human-in-the-loop interventions, ensures adherence to predefined plans.
Disadvantages: Supervisor context becomes bottleneck, supervisor failures cascade to all workers, "telephone game" problem where supervisors paraphrase sub-agent responses incorrectly.
The Telephone Game Problem and Solution LangGraph benchmarks found supervisor architectures initially performed 50% worse than optimized versions due to the "telephone game" problem where supervisors paraphrase sub-agent responses incorrectly, losing fidelity.
The fix: implement a forward_message tool allowing sub-agents to pass responses directly to users:
def forward_message(message: str, to_user: bool = True):
"""
Forward sub-agent response directly to user without supervisor synthesis.
Use when:
- Sub-agent response is final and complete
- Supervisor synthesis would lose important details
- Response format must be preserved exactly
"""
if to_user:
return {"type": "direct_response", "content": message}
return {"type": "supervisor_input", "content": message}
With this pattern, swarm architectures slightly outperform supervisors because sub-agents respond directly to users, eliminating translation errors.
Implementation note: Implement direct pass-through mechanisms allowing sub-agents to pass responses directly to users rather than through supervisor synthesis when appropriate.
Pattern 2: Peer-to-Peer/Swarm The peer-to-peer pattern removes central control, allowing agents to communicate directly based on predefined protocols. Any agent can transfer control to any other through explicit handoff mechanisms.
def transfer_to_agent_b():
return agent_b # Handoff via function return
agent_a = Agent(
name="Agent A",
functions=[transfer_to_agent_b]
)
When to use: Tasks requiring flexible exploration, tasks where rigid planning is counterproductive, tasks with emergent requirements that defy upfront decomposition.
Advantages: No single point of failure, scales effectively for breadth-first exploration, enables emergent problem-solving behaviors.
Disadvantages: Coordination complexity increases with agent count, risk of divergence without central state keeper, requires robust convergence constraints.
Implementation note: Define explicit handoff protocols with state passing. Ensure agents can communicate their context needs to receiving agents.
Pattern 3: Hierarchical Hierarchical structures organize agents into layers of abstraction: strategic, planning, and execution layers. Strategy layer agents define goals and constraints; planning layer agents break goals into actionable plans; execution layer agents perform atomic tasks.
Strategy Layer (Goal Definition) -> Planning Layer (Task Decomposition) -> Execution Layer (Atomic Tasks)
When to use: Large-scale projects with clear hierarchical structure, enterprise workflows with management layers, tasks requiring both high-level planning and detailed execution.
Advantages: Mirrors organizational structures, clear separation of concerns, enables different context structures at different levels.
Disadvantages: Coordination overhead between layers, potential for misalignment between strategy and execution, complex error propagation.
Context Isolation as Design Principle
The primary purpose of multi-agent architectures is context isolation. Each sub-agent operates in a clean context window focused on its subtask without carrying accumulated context from other subtasks.
Isolation Mechanisms Full context delegation: For complex tasks where the sub-agent needs complete understanding, the planner shares its entire context. The sub-agent has its own tools and instructions but receives full context for its decisions.
Instruction passing: For simple, well-defined subtasks, the planner creates instructions via function call. The sub-agent receives only the instructions needed for its specific task.
File system memory: For complex tasks requiring shared state, agents read and write to persistent storage. The file system serves as the coordination mechanism, avoiding context bloat from shared state passing.
Isolation Trade-offs Full context delegation provides maximum capability but defeats the purpose of sub-agents. Instruction passing maintains isolation but limits sub-agent flexibility. File system memory enables shared state without context passing but introduces latency and consistency challenges.
The right choice depends on task complexity, coordination needs, and acceptable latency.
Consensus and Coordination
The Voting Problem Simple majority voting treats hallucinations from weak models as equal to reasoning from strong models. Without intervention, multi-agent discussions devolve into consensus on false premises due to inherent bias toward agreement.
Weighted Voting Weight agent votes by confidence or expertise. Agents with higher confidence or domain expertise carry more weight in final decisions.
Debate Protocols Debate protocols require agents to critique each other's outputs over multiple rounds. Adversarial critique often yields higher accuracy on complex reasoning than collaborative consensus.
Trigger-Based Intervention Monitor multi-agent interactions for specific behavioral markers. Stall triggers activate when discussions make no progress. Sycophancy triggers detect when agents mimic each other's answers without unique reasoning.
Framework Considerations
Different frameworks implement these patterns with different philosophies. LangGraph uses graph-based state machines with explicit nodes and edges. AutoGen uses conversational/event-driven patterns with GroupChat. CrewAI uses role-based process flows with hierarchical crew structures.
Anti-Patterns
| Anti-Pattern | Problem | Solution |
|---|---|---|
| Role-based decomposition instead of context-based | Agents divided by job title ("researcher", "writer") rather than by context isolation needs | Decompose by context boundary: each sub-agent handles a distinct context scope, not an org-chart role |
| Supervisor paraphrasing all sub-agent outputs | "Telephone game" — fidelity loss when supervisor summarizes | Implement forward_message for direct pass-through; supervisor only synthesizes when aggregation is needed |
| No time-to-live limits on agents | Agents run indefinitely, burning tokens without convergence | Set TTL per agent execution; enforce convergence checks |
| Skipping output validation between agents | Errors propagate silently to downstream agents | Validate outputs before passing to consumers; use schema constraints |
| Simple majority voting without weighting | Weak model hallucinations count equally with strong model reasoning | Use weighted voting by confidence or expertise; use debate protocols for high-stakes decisions |
| Full context delegation everywhere | Negates the purpose of multi-agent (context isolation) | Use instruction passing for well-defined subtasks; reserve full delegation for complex decisions |
| Choosing pattern by organizational metaphor | "Our team has a manager so we need a supervisor" — wrong basis | Choose by coordination needs: supervisor for control, swarm for exploration, hierarchical for layered abstraction |
| Ignoring token cost multiplier | 15× token cost surprises teams at billing time | Budget tokens explicitly; monitor per-agent usage; compare single-agent baseline |
Examples
Example 1: Research Team Architecture
Supervisor
├── Researcher (web search, document retrieval)
├── Analyzer (data analysis, statistics)
├── Fact-checker (verification, validation)
└── Writer (report generation, formatting)
Example 2: Handoff Protocol
def handle_customer_request(request):
if request.type == "billing":
return transfer_to(billing_agent)
elif request.type == "technical":
return transfer_to(technical_agent)
elif request.type == "sales":
return transfer_to(sales_agent)
else:
return handle_general(request)
Guidelines
- Design for context isolation as the primary benefit of multi-agent systems
- Choose architecture pattern based on coordination needs, not organizational metaphor
- Implement explicit handoff protocols with state passing
- Use weighted voting or debate protocols for consensus
- Monitor for supervisor bottlenecks and implement checkpointing
- Validate outputs before passing between agents
- Set time-to-live limits to prevent infinite loops
- Test failure scenarios explicitly
Integration
This skill builds on context-fundamentals and context-degradation. It connects to:
- memory-systems - Shared state management across agents
- tool-design - Tool specialization per agent
- context-optimization - Context partitioning strategies
References
Internal reference:
- Frameworks Reference - Detailed framework implementation patterns
Related skills in this collection:
- context-fundamentals - Context basics
- memory-systems - Cross-agent memory
- context-optimization - Partitioning strategies
External resources:
- LangGraph Documentation - Multi-agent patterns and state management
- AutoGen Framework - GroupChat and conversational patterns
- CrewAI Documentation - Hierarchical agent processes
- Research on Multi-Agent Coordination - Survey of multi-agent systems