# SafeSieve: From Heuristics to Experience in Progressive Pruning for LLM-based Multi-Agent Communication

arXiv: https://arxiv.org/abs/2508.11733 · HTML: https://mlsanity.com/paper/2508.11733/

## Key facts

- Cited by 12 papers in the map (data through 14 Oct 2026).
- Placed in General LLM agent frameworks and benchmarks, one of 16 directions in LLM agents and tool use.
- Builds on 12 of its 34 references that are in the map.
- First posted on arXiv 15 Aug 2025.

## Placed in

- LLM agents and tool use / [General LLM agent frameworks and benchmarks](https://mlsanity.com/agents/general-llm-agent-frameworks-and-benchmarks/)

## Builds on (most cited first, 12 of 12)

- [Let's Verify Step by Step](https://mlsanity.com/paper/2305.20050/) (2023)
- [Lost in the Middle: How Language Models Use Long Contexts](https://mlsanity.com/paper/2307.03172/) (2023)
- [ChatDev: Communicative Agents for Software Development](https://mlsanity.com/paper/2307.07924/) (2023)
- [Are NLP Models really able to Solve Simple Math Word Problems?](https://mlsanity.com/paper/2103.07191/) (2021)
- [Program Induction by Rationale Generation : Learning to Solve and Explain Algebraic Word Problems](https://mlsanity.com/paper/1705.04146/) (2017)
- [A Dynamic LLM-Powered Agent Network for Task-Oriented Agent Collaboration](https://mlsanity.com/paper/2310.02170/) (2023)
- [Chain of Agents: Large Language Models Collaborating on Long-Context Tasks](https://mlsanity.com/paper/2406.02818/) (2024)
- [A Comprehensive Survey in LLM(-Agent) Full Stack Safety: Data, Training and Deployment](https://mlsanity.com/paper/2504.15585/) (2025)
- [Embodied LLM Agents Learn to Cooperate in Organized Teams](https://mlsanity.com/paper/2403.12482/) (2024)
- [Self-Evolving Multi-Agent Collaboration Networks for Software Development](https://mlsanity.com/paper/2410.16946/) (2024)
- [Adaptive Graph Pruning for Multi-Agent Communication](https://mlsanity.com/paper/2506.02951/) (2025)
- [AnyMAC: Cascading Flexible Multi-Agent Collaboration via Next-Agent Prediction](https://mlsanity.com/paper/2506.17784/) (2025)

## Built on by (most cited first, 12 of 12)

- [Toward Efficient Agents: Memory, Tool learning, and Planning](https://mlsanity.com/paper/2601.14192/) (2026)
- [Stop Wasting Your Tokens: Towards Efficient Runtime Multi-Agent Systems](https://mlsanity.com/paper/2510.26585/) (2025)
- [Collab-REC: An LLM-based Agentic Framework for Balancing Recommendations in Tourism](https://mlsanity.com/paper/2508.15030/) (2025)
- [TodyComm: Task-Oriented Dynamic Communication for Multi-Round LLM-based Multi-Agent System](https://mlsanity.com/paper/2602.03688/) (2026)
- [Self-Evolving Agents as Dynamic Graph Transformation: A Survey and New Perspective](https://mlsanity.com/paper/2608.18104/) (2026)
- [ST-EVO: Towards Generative Spatio-Temporal Evolution of Multi-Agent Communication Topologies](https://mlsanity.com/paper/2602.14681/) (2026)
- [AgentSlimming: Towards Efficient and Cost-Aware Multi-Agent Systems](https://mlsanity.com/paper/2605.08813/) (2026)
- [Differentiable Mixture-of-Agents Incentivizes Swarm Intelligence of Large Language Models](https://mlsanity.com/paper/2605.15706/) (2026)
- [SoK: When Safe Agents Fail Together: The Security of Multi Agent LLM Systems](https://mlsanity.com/paper/2609.00595/) (2026)
- [BANDMAS: Causality-Inspired Semantic Packet Scheduling for Bandwidth-Efficient Multi-Agent Collaboration](https://mlsanity.com/paper/2608.00458/) (2026)
- [Privacy-Preserving Topology-Guided Safety for LLM-Based Multi-Agent Systems via Federated Graph Learning](https://mlsanity.com/paper/2609.02967/) (2026)
- [Social Circuits behind Multi-agent Echo Chambers](https://mlsanity.com/paper/2609.34444/) (2026)

## Abstract (arXiv)

LLM-based multi-agent systems exhibit strong collaborative capabilities but often suffer from redundant communication and excessive token overhead. Existing methods typically enhance efficiency through pretrained GNNs or greedy algorithms, but often isolate pre- and post-task optimization, lacking a unified strategy. To this end, we present SafeSieve, a progressive and adaptive multi-agent pruning algorithm that dynamically refines the inter-agent communication through a novel dual-mechanism. SafeSieve integrates initial LLM-based semantic evaluation with accumulated performance feedback, enabling a smooth transition from heuristic initialization to experience-driven refinement. Unlike existing greedy Top-k pruning methods, SafeSieve employs 0-extension clustering to preserve structurally coherent agent groups while eliminating ineffective links. Experiments across benchmarks (SVAMP, HumanEval, etc.) showcase that SafeSieve achieves 94.01% average accuracy while reducing token usage by 12.4%-27.8%. Results further demonstrate robustness under prompt injection attacks (1.23% average accuracy drop). In heterogeneous settings, SafeSieve reduces deployment costs by 13.3% while maintaining performance. These results establish SafeSieve as an efficient, GPU-free, and scalable framework for practical multi-agent systems. Our code can be found here: https://github.com/csgen/SafeSieve