mlsanityData to 7 Oct 2026

Roles with Rails: Contract-Preserving Role Evolution in Multi-Agent Structured Reasoning

arXiv 2605.28433 ↗First posted 27 May 2026cs.CLcited by 0

On the map: LLM reasoning → Reasoning for agents, planning and code

also in 1 more
Abstract (arXiv)

Role-based LLM multi-agent systems need adaptive role pools, yet adapting such systems is not merely a matter of prompt optimization: roles often carry structural obligations, including capability coverage, message compatibility, validation, final-answer aggregation, and parser-compatible output protocols. Existing systems either fix the role inventory and lose adaptivity, or allow unconstrained generation to induce role drift, removing structurally necessary roles and breaking answer contracts. We formulate this as contract-preserving role evolution, requiring every committed edit to preserve five structural contracts (capability, communication, validation, aggregation, output protocol). We instantiate this formulation in SERO, a Self-Evolving Role Orchestration framework that evolves a typed role-card pool through credit-guided retrieval, a credit-ranked communication DAG with a protected terminal aggregator and conditional validator repair, and a contextual-bandit controller whose LLM-proposed edits are committed only when they preserve the contracts and improve task score. Experiments on real-world reasoning benchmarks across three LLM backbones confirm the value of contract-preserving role evolution.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in Reasoning for agents, planning and code, one of 17 directions in LLM reasoning. Builds on 14 of its 38 references that are in the map. First posted on arXiv 27 May 2026.

Builds on

14 of 38 references.

Closest ancestors · 2

Same direction, newest first

  1. 2026MetaGen: Self-Evolving Roles and Topologies for Multi-Agent LLM Reasoning27 Jan
  2. 2025Multi-Agent Evolve: LLM Self-Improve through Co-evolution27 Oct

Other references · 12

  1. 2025Qwen3 Technical Reportcited by 8,195
  2. 2022Self-Consistency Improves Chain of Thought Reasoning in Language Modelscited by 5,601
  3. 2023AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversationcited by 1,906
  4. 2023A Dynamic LLM-Powered Agent Network for Task-Oriented Agent Collaborationcited by 281
  5. 2024NATURAL PLAN: Benchmarking LLMs on Natural Language Planningcited by 125
  6. 2024Exponential Moving Average of Weights in Deep Learning: Dynamics and Benefitscited by 65
  7. 2025Adaptive Graph Pruning for Multi-Agent Communicationcited by 28
  8. 2025Leveraging Large Language Models for Effective and Explainable Multi-Agent Credit Assignmentcited by 17
  9. 2026Dynamic Role Assignment for Multi-Agent Debatecited by 5
  10. 2026TodyComm: Task-Oriented Dynamic Communication for Multi-Round LLM-based Multi-Agent Systemcited by 5
  11. 2026GoAgent: Group-of-Agents Communication Topology Generation for LLM-based Multi-Agent Systemscited by 5
  12. 2026SC-MAS: Constructing Cost-Efficient Multi-Agent Systems with Edge-Level Heterogeneous Collaborationcited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026At Equal Inference Cost, Multi-Agent Structure Does Not Beat a Single Frozen Agent
  2. 2026Role-Agent: Bootstrapping LLM Agents via Dual-Role Evolution
  3. 2026ExRole: From Team Trajectories to Executable Roles in Multi-Agent Language Models
  4. 2026Unifying Temporal and Structural Credit Assignment in LLM-Based Multi-Agent Prompt Optimization
  5. 2026HarnessForge: Joint Harness and Policy Evolution for Adaptive Agent Systems
  6. 2026TacoMAS: Test-Time Co-Evolution of Topology and Capability in LLM-based Multi-Agent Systems
  7. 2026From Task Solving to Robust Real-World Adaptation in LLM Agents
  8. 2026Unified Context Evolution for LLM Agents