Decomposition of complex tasks into a set of autonomous agents with specialized roles that communicate and coordinate, enabling systems to exceed the reasoning limitations of single-model inference and complete long-horizon tasks that cannot be completed within a single LLM call.
Core computational and decision-making unit of the system.
Autonomous computational entity with its own internal state, perception, reasoning, and action capabilities. In LLM-based MAS: a language model with a system prompt defining the agent's role, goal, and constraints.
Enables coordination and state transfer between agents.
Mechanism for information exchange between agents. Can take the form of direct message passing, shared memory, publish-subscribe systems, or event queues.
Managing global task progress and coordinating between agents.
Component responsible for workflow management: task decomposition, routing to agents, dependency management, error handling, and aggregating final results. Can be an LLM agent or a programmatic controller.
Persists state across agent calls and system sessions.
State storage mechanisms: short-term memory (conversation context / LLM context window), long-term memory (vector database, external database), episodic memory (interaction history), and procedural memory (learned procedures).
Extends agent capabilities beyond language processing to actions in the external world.
Layer integrating agents with external systems: APIs, search engines, code interpreters, databases, file services. Enables agents to act beyond their textual context.
Conditionally parallel
MAS systems can be fully parallel (agents operating asynchronously on independent subtasks), partially sequential (with synchronization barriers), or hybrid, depending on the task topology and its dependency graph.
Conditional
Input dependent
Not all agents are active for every task — the set of active agents depends on the nature of the task and the orchestrator's dynamic routing.
Number of Agents
Number of agents in the system. Higher agent count can increase parallelism and specialization but complicates coordination and increases cost.
Coordination Topology
Communication and dependency pattern between agents: sequential chain, hierarchy (orchestrator-worker), peer-to-peer mesh, publish-subscribe, parallel fan-out/fan-in, or hybrid DAG.
Memory Architecture
Type and scope of memory allocated to each agent and the system as a whole: context-only (within LLM context window), external long-term (vector database), shared across agents, or private per agent.
Human Involvement
Mode and entry points for human participation: none (full autonomy), approval at every step, approval at key decisions, or intervention only on errors.
Agent specialization degree
Degree to which individual agents have specialized roles (e.g., search agent, coding agent, verification agent) versus general-purpose agents.
Incorrect output from one agent is passed as input to the next, leading to error accumulation across the pipeline. Without a validator agent, this can result in completely incorrect final outputs.
Deploy critic/validator agents after key steps; use multiple independent execution paths with voting or result aggregation.
Each inter-agent communication consumes tokens (passing conversation history, context, tools). With many agents and long communication chains, costs can grow disproportionately fast.
Use context compression; limit the history passed between agents to the necessary minimum; use lighter models for simpler subtasks.
Agents can enter loops — e.g., one agent requests revision from another, which requests clarification back — without a termination mechanism. Absence of stopping criteria causes infinite loops.
Define explicit termination criteria (maximum number of iterations, a 'TERMINATE' condition, timeout); use a supervising monitor agent.
During parallel execution, agents may operate on inconsistent versions of shared state, leading to conflicts and race conditions, especially when shared memory is non-transactional.
Use transactional updates for shared state; design agents to be idempotent; minimize write-write contention by assigning distinct state partitions to separate agents.
Agents with poorly defined roles may duplicate work, conflict in decision-making, or skip tasks because each assumes another agent will handle it.
Precisely define each agent's scope of responsibility in the system prompt; apply formal handoff protocols and task-receipt acknowledgments.
Wooldridge and Jennings formalize the notion of intelligent agents and MAS
breakthroughWooldridge and Jennings publish 'Intelligent Agents: Theory and Practice' (Knowledge Engineering Review), defining agent properties (autonomy, reactivity, pro-activeness, social ability) and foundations of MAS theory.
Gerhard Weiss edits 'Multiagent Systems: A Modern Approach to Distributed Artificial Intelligence' (MIT Press)
Textbook standardizing MAS terminology and architecture, becoming the primary academic reference for the following decade.
LLM-MAS framework explosion: CAMEL, AutoGen, MetaGPT
breakthrough2023 sees the first wave of LLM-based MAS frameworks: CAMEL (Li et al., March 2023), AutoGen (Wu et al., Microsoft, August 2023), MetaGPT (Hong et al., August 2023), transforming MAS from rule-based to language-model-based systems.
Standardization of communication protocols (MCP, A2A) and orchestration frameworks (LangGraph, CrewAI)
Anthropic announces Model Context Protocol (MCP), Google announces Agent-to-Agent Protocol (A2A). More mature frameworks emerge: LangGraph (LangChain), CrewAI, supporting complex agent topologies with state persistence and controlled flow.
MAS is an architectural pattern operating above the hardware layer. Individual agents may run on GPUs (LLM inference), CPUs (orchestration logic), or in the cloud, depending on the implementation.
For LLM-MAS systems using large models, the hardware bottleneck becomes GPU memory bandwidth during parallel inference when multiple agents share the same model.
| Title | Publisher | Type |
|---|---|---|
| Multi-Agent Systems: An Introduction | — | — |
