Mind Your Own Role: Reducing Role Overstepping for Effective and Efficient Multi-Agent Collaboration via Dynamic Role Activation Steering

Xingke Yang; Sicong Li; Lingfeng Yao; Rui Chen; Gen Li; Hao Wang; Xin Fu; Miao Pan

cs.MAOct 11, 2026

Mind Your Own Role: Reducing Role Overstepping for Effective and Efficient Multi-Agent Collaboration via Dynamic Role Activation Steering

Authors: Xingke Yang, Sicong Li, Lingfeng Yao, Rui Chen, Gen Li, Hao Wang, Xin Fu, Miao Pan

Organizations: University of Houston, USA · Independent Researcher, USA · Stevens Institute of Technology, USA

Abstract

Multi-Agent Systems (MAS) have shown strong capability in solving complex tasks by assigning different agents specialized roles and corresponding subtasks. However, agents may overstep their assigned responsibilities and perform subtasks belonging to other roles, which can disrupt workflow, reduce system efficiency, and degrade task performance. To address this issue, we investigate role-related patterns in the model's internal activation space and uncover two key observations: (i) role instructions induce consistent directional shifts in activation space, and (ii) role-adherent and role-overstepping trajectories exhibit distinct activation patterns. Based on these observations, we propose Dynamic Role Activation Steering (DRAS), a lightweight and effective inference-time activation intervention approach to reduce role overstepping in multi-agent collaboration. Specifically, DRAS constructs a role steering vector and a lightweight role-overstepping detector for each role. During LLM generation, the detector estimates the probability of role overstepping from the current activation state, and DRAS dynamically adjusts the steering strength accordingly to guide the agent toward its assigned responsibilities. Experimental results show that DRAS effectively reduces the number of role-overstepping trajectories by up to 86.6% compared with Vanilla MAS, thereby improving task quality score by 10.0%--11.1%, achieving a 1.50 $\times$ --1.61 $\times$ task-completion speedup, and reducing token consumption by 26.8%--38.7%.

Figures & tables

Panel of Figure 2 — Figure 2: PCA projections of internal activations in ChatDev for the CEO role, extracted at selected layers of Qwen2.5-7B and Llama-3.1-8B.

Figure 3: Role overstepping in multi-agent collaboration and our proposed Dynamic Role Activation Steering (DRAS). During generation, DRAS first detects the role-overstepping probability from the current activation state and then dynamically adjusts the steering strength to guide the agent toward its assigned role.

Method	ChatDev-Qwen			MetaGPT-Llama
Method	RO Rate $\downarrow$	RO Traj. # $\downarrow$	RC Score $\uparrow$	RO Rate $\downarrow$	RO Traj. # $\downarrow$	RC Score $\uparrow$
Vanilla MAS	42.67	987	0.621	25.33	423	0.784
Few-shot	25.92	389	0.713	18.75	261	0.786
Post-hoc	15.25	257	0.725	15.33	203	0.813
LoRA FT	11.50	196	0.798	19.33	247	0.834
DRAS (Ours)	7.75	132	0.832	7.83	114	0.855

Table 1: Role-overstepping mitigation performance of different methods across multi-agent systems and LLM backbones on SRDD. ( RO Rate : Role-Overstepping Rate; RO Traj. # : Number of Role Overstepping Trajectories; RC Score : Role Clarity Score)

System & Model	Method	Task Performance				Efficiency
System & Model	Method	Compl. $\uparrow$	Exec. $\uparrow$	Cons. $\uparrow$	Qual. $\uparrow$	Compl. Time (s) $\downarrow$ / (Speedup) $\uparrow$	Tokens $\downarrow$
ChatDev- Qwen	Vanilla MAS	0.7017	0.8708	0.693	0.7552	341.3 (1.00 $\times$ )	37.7K
	Few-shot	0.7133	0.9292	0.795	0.8125	293.8 (1.16 $\times$ )	31.9K
	Post-hoc	0.7225	0.9375	0.774	0.8113	461.0 (0.74 $\times$ )	51.3K
	LoRA FT	0.7267	0.9358	0.828	0.8302	227.9 (1.50 $\times$ )	26.1K
	DRAS (Ours)	0.7408	0.955	0.819	0.8383	211.4 (1.61 $\times$ )	24.3K

Table 2: End-to-end task performance and efficiency of different methods across multi-agent systems and LLM backbones on SRDD. ( Compl. , Exec. , Cons. , and Qual. denote Completeness, Executability, Consistency, and Quality, respectively.)

Method	Peak Mem. (GB) $\downarrow$	Computation (PFLOPs) $\downarrow$
LoRA FT	45.3	374.0
DRAS	16.4	42.9

Table 3: Offline preparation cost per role on Qwen2.5-7B.

System	Method	RO Rate $\downarrow$	RO Traj. # $\downarrow$	RC Score $\uparrow$	Quality $\uparrow$	Compl. Time (s) $\downarrow$	Tokens $\downarrow$
ChatDev- Qwen	Static Steering	5.92	93	0.807	0.8207	206.73	22.8K
	Binary Gating	9.83	151	0.819	0.8223	237.1	27.6K
	DRAS	7.75	132	0.832	0.8383	211.4	24.3K
MetaGPT- Llama	Static Steering	6.75	92	0.818	0.7422	309.1	28.7K
	Binary Gating	10.50	147	0.847	0.7512	324.7	31.4K
	DRAS	7.83	114	0.855	0.7608	318.5	29.3K

Table 4: Ablation study of dynamic steering. Static Steering applies a fixed steering strength at each decoding step, Binary Gating converts the detector probability into a binary on/off decision using a threshold, and DRAS dynamically adjusts the steering strength according to the predicted role-overstepping probability.

Panel of Figure 4 — Figure 4: Layer-wise steering and detection performance and sensitivity to the base steering strength in DRAS. Results are reported on the SWE-Dev validation set for the CEO and CPO roles in ChatDev using Qwen2.5-7B and Llama-3.1-8B.

Figure 5: A role-overstepping trajectory and its DRAS-corrected trajectory for the Code Reviewer role in ChatDev-Qwen. The Code Reviewer is instructed to only provide high-priority comments and revision suggestions, and red boxes highlight role-overstepping content. Token colors indicate the role-overstepping probability predicted by the detector.

Appendix figures & tables8 assets

Supplementary material from the paper’s appendix.

Appendix

Role	Agreement (%)
CEO	100
CPO	98
Code Reviewer	96

Appendix

Table 5: Agreement between GPT-4o and manual annotations on ChatDev-Qwen.

Panel of Figure 6 — Figure 6: Prompt used for GPT-4o role-overstepping evaluation.

Framework	Role	RO Rate (%)	Selected
ChatDev	CEO	35.70	$\checkmark$
	CPO	31.90	$\checkmark$
	CTO	1.20
	Programmer	1.40
	Code Reviewer	18.50	$\checkmark$
	Software Test Engineer	0.70

Appendix

Table 6: Role-overstepping prevalence across different roles in ChatDev and MetaGPT on 2,000 SWE-Dev tasks using Qwen2.5-7B-Instruct. Checkmarks indicate the selected target roles for role-overstepping mitigation.

Panel of Figure 7 — Figure 7: Prompt augmentation used for the Few-shot method.

Panel of Figure 8 — Figure 8: Modified output instruction used in the Post-hoc method.

Panel of Figure 9 — Figure 9: Additional instruction appended to the original agent prompt for response regeneration in the Post-hoc method.

Figure 10: Offline preparation of DRAS, including role steering vector construction and role-overstepping detector training.

Configuration	Qwen2.5-7B	Llama-3.1-8B
Intervention layer	18	22
Base steering strength $\alpha$	1.2	0.8

Appendix

Table 7: Implementation configurations of DRAS for different backbone LLMs.

Multi-Agent LLM Systems

Multi-agent large language model systems coordinate multiple AI agents to solve tasks through communication, specialization, shared tools, and memory.

See all 1,001 papers

Improving Role Consistency in Multi-Agent Collaboration via Quantitative Role Clarity
Apr 3, 2026Guoling Zhou, Wenpei Han, Fengqin Yang +3Behavioral Consistency of LLM Agents LLM Agents
School of Information Science and Technology, Northeast Normal University, Changchun, Jilin · School of Computer Science and Engineering, Guangxi Normal University, Guilin, Guangxi
Token-Efficient Multi-Agent Collaboration via System One-Guided Computational Division of Labor
Oct 6, 2026Zihan Zhou, Xinzhe Hu, Hanxu Yang +2Multi-Agent LLM Systems Multi-Agent System Optimization
University of Electronic Science and Technology of China · Southwestern University of Finance and Economics
One Model, Many Minds: Unlocking Multi-Agent Synergy in a Single Agent via Mixture of Roles
Aug 27, 2026Zhichen Zeng, Huiyuan Chen, Jingru Cheng +7Language Model Steering Multi-Agent Reasoning
University of Illinois Urbana-Champaign · Amazon

Apr 3, 2026J/K move · Enter open · S save

Improving Role Consistency in Multi-Agent Collaboration via Quantitative Role Clarity

Guoling Zhou, Wenpei Han, Fengqin Yang, Li Wang+2

School of Information Science and Technology, Northeast Normal University, Changchun, Jilin · School of Computer Science and Engineering, Guangxi Normal University, Guilin, Guangxi

Behavioral Consistency of LLM Agents LLM Agents Multi-Agent Collaboration LLM Agent Training LLM Agent Reliability

Open paper AskPDF arXiv

In large language model (LLM)-driven multi-agent systems, disobey role specification (failure to adhere to the defined responsibilities and constraints of an assigned role, potentially leading to an agent behaving like another) is a major failure mode \cite{DBLP:journals/corr/abs-2503-13657}. To address this issue, in the present paper, we propose a quantitative role clarity to improve role consistency. Firstly, we construct a role assignment matrix

S(φ)=[s_{ij}(φ)]

, where

s_{ij}(φ)

is the semantic similarity between the

i

-th agent's behavior trajectory and the

j

-th agent's role description. Then we define role clarity matrix

M(φ)

\text{softmax}(S(φ))-I

, where

\text{softmax}(S(φ))

is a row-wise softmax of

S(φ)

and

I

is the identity matrix. The Frobenius norm of

M(φ)

quantifies the alignment between agents' role descriptions and their behaviors trajectory. Moreover, we employ the role clarity matrix as a regularizer during lightweight fine-tuning to improve role consistency, thereby improving end-to-end task performance. Experiments on the ChatDev multi-agent system show that our method substantially improves role consistency and task performance: with Qwen and Llama, the role overstepping rate decreases from

46.4\%

8.4\%

and from

43.4\%

0.2\%

, respectively, and the role clarity score increases from

0.5328

0.9097

and from

0.5007

0.8530

, respectively, the task success rate increases from

0.6769

0.6909

and from

0.6174

0.6763

, respectively.

Abstract

Multi-Agent LLM Systems

Explore similar work

Improving Role Consistency in Multi-Agent Collaboration via Quantitative Role Clarity

Token-Efficient Multi-Agent Collaboration via System One-Guided Computational Division of Labor

One Model, Many Minds: Unlocking Multi-Agent Synergy in a Single Agent via Mixture of Roles