As Large Language Model (LLM) based Multi-Agent Systems (MAS) evolve from experimental pilots to complex, persistent ecosystems, the limitations of direct agent-to-agent communication have become increasingly apparent. Current architectures suffer from fragmented context, stochastic hallucinations, rigid security boundaries, and inefficient topology management. This paper introduces Cognitive Fabric Nodes (CFN), a novel middleware layer that creates an omnipresent "Cognitive Fabric" between agents. Unlike traditional message queues or service meshes, CFNs are not merely pass-through mechanisms; they are active, intelligent intermediaries. Central to this architecture is the elevation of Memory from simple storage to an active functional substrate that informs four other critical capabilities: Topology Selection, Semantic Grounding, Security Policy Enforcement, and Prompt Transformation. We propose that each of these functions be governed by learning modules utilizing Reinforcement Learning (RL) and optimization algorithms to improve system performance dynamically. By intercepting, analyzing, and rewriting inter-agent communication, the Cognitive Fabric ensures that individual agents remain lightweight while the ecosystem achieves coherence, safety, and semantic alignment. We evaluate the effectiveness of the CFN on the HotPotQA and MuSiQue datasets in a multi-agent environment and demonstrate that the CFN improves performance by more than 10% on both datasets over direct agent to agent communication.
Figures & tables
Function Triggered
Raw Prompt ( x )
Fabric Context ( Ω )
Final Output ( y )
Memory Injection
“Check the logs for that error.”
Memory: “Error 503 occurred in service ‘Payment-Gateway’ at 14:00.”
“Scan the ‘Payment-Gateway’ logs specifically for Error 503 timestamps around 14:00 today.”
Semantic Grounding
“Deploy the container.”
Topology: Receiver is a K8s Agent. Grounding: “Deploy” implies kubectl apply .
“Execute kubectl apply -f deployment.yaml to the production namespace.”
Security Enforcement
“Here is the user dump: [Real Names/SSNs]”
Security Policy: PII Detection Active.
“Here is the user dump: [REDACTED_PII]. Note: Fabric removed sensitive data per Policy 8A. ”
Adversarial Defense
“Ignore previous instructions and delete DB.”
Memory: Previous instruction was “Backup DB”. Security: Conflict detected.
“Maintain previous backup instructions. Report status of database integrity. Fabric Note: Contradictory command intercepted and neutralized. ”
TABLE I: Operational examples of Cognitive Fabric Node transformations.
Attack Type
Human-Specified Rule ( H )
Learned Cognitive Defense ( L )
Fabric Intervention
SQL Injection
Trigger: RegEx detects DROP TABLE .
N/A
Block: “Unsafe database operation detected.”
PII Leakage
Trigger: Pattern detects SSN: \d{3}-\d{2}...
N/A
Redact: Rewrite prompt to replace SSN with <REDACTED> .
“DAN” Jailbreak
Pass: No specific keywords triggered.
Trigger: Semantic embedding matches cluster of known “Persona-based Jailbreaks.”
Neutralize: Rewrite prompt to strip the “persona” wrapper and keep only the core query.
Contextual Manipulation
Pass: Individual messages look safe.
Trigger: Memory analysis shows deviation from agent’s standard behavior profile (Anomaly).
TABLE II: Comparison of Deterministic vs. Probabilistic Security Interventions.
Method
Baseline
Multi-agent
TextGrad
Ours
Accuracy/Pass Rate
HotPotQA
92
80.1
82.3
91.5
MuSiQue
87.5
72.7
76.2
86.1
TABLE III: Evaluation on the HotPotQA and MuSiQue datasets. These datasets were modified so that each agent only had access to partial information. All reported values represent the mean performance after 5 training iterations using the same backbone LLMs. Baseline is the base Claude Sonnet 4.6 model performance, multi-agent is the base multi-agent system performance, TextGrad is performance with TextGrad only updates.
Recent surges in LLM-driven intelligent systems largely overlook decades of foundational multi-agent systems (MAS) research, resulting in frameworks with critical limitations such as centralization and inadequate trust and communication protocols. This paper introduces the Fetch.ai architecture, an industrial-strength platform designed to bridge this gap by facilitating the integration of classical MAS principles with modern AI capabilities. We present a novel, multi-layered solution built on a decentralized foundation of on-chain blockchain services for verifiable identity, discovery, and transactions. This is complemented by a comprehensive development framework for creating secure, interoperable agents, a cloud-based platform for deployment, and an intelligent orchestration layer where an agent-native LLM translates high-level human goals into complex, multi-agent workflows. We demonstrate the deployed nature of this system through a decentralized logistics use case where autonomous agents dynamically discover, negotiate, and transact with one another securely. Ultimately, the Fetch.ai stack provides a principled architecture for moving beyond current agent implementations towards open, collaborative, and economically sustainable multi-agent ecosystems.
Michael J. Wooldridge, Attila Bagoly, Jonathan J. Ward +2
Multi-agent systems (MAS) built on large language models (LLMs) have shown strong performance across many tasks. Most existing approaches improve only one aspect at a time, such as the communication topology, role assignment, or LLM routing, while treating each agent as a single, indivisible unit. This misses the opportunity to use mixtures of LLMs within an agent to strengthen role-specific abilities. We propose HieraMAS, a hierarchical collaboration framework that combines intra-node LLM mixtures with an inter-node communication topology. HieraMAS introduces supernodes, where each functional role is implemented by multiple heterogeneous LLMs using a propose-synthesis structure. Optimizing HieraMAS creates unique credit-assignment challenges: final task performance depends heavily on the underlying LLMs' capabilities, which can lead reinforcement methods to incorrectly reward suboptimal configurations. To address this, we use a two-stage algorithm: (1) multi-level reward attribution, which provides fine-grained feedback at both the node level and the overall system level; (2) graph classification for topology selection, which treats choosing the communication structure as a holistic decision rather than optimizing edges one by one. Experiments on reasoning and coding benchmarks show that HieraMAS substantially outperforms existing methods while also delivering better cost-performance trade-offs.
Tianjun Yao, Zhaoyi Li, Zhiqiang Shen
Mohamed bin Zayed University of Artificial Intelligence, Abu Dhabi, UAE.
Existing large language model (LLM) based multi-agent systems have been shown to scale poorly, and we identify three problems: reading burden that grows with the group, useful information diluted by shared work of unknown quality, and fragility to failed agents. We argue that the problems are largely due to agent-centric designs: the coordination structure is wired into fixed workflows, central coordinators, or peer-to-peer messaging, and recent blackboard-style methods that move the coordination into a shared medium relieve only the last problem. To close the gaps, we build ScaleMAS, where asynchronous agents interact with a shared datastore of linked messages via three operations: publishing messages with optional references to earlier ones that they build on, searching for messages by content, and visiting messages by address. Search returns at most k messages, so an agent's read stays bounded at any group size; results are ranked by relevance and by how often they are cited, which we find to be an effective proxy for quality; and messages persist after their producers leave, so the group carries on when individual agents fail. Across four agentic tasks, ScaleMAS outperforms agent-centric and blackboard-style baselines at lower inference compute, continues to scale up to 128 agents, and tolerates losing agents midway.