DIG to Heal: Scaling General-purpose Agent Collaboration via Explainable Dynamic Decision Paths

Hanqing Yang; Hyungwoo Lee; Yuhang Yao; Zhiwei Liu; Kay Liu; Jingdi Chen; Carlee Joe-Wong

cs.AIFeb 27, 2026

DIG to Heal: Scaling General-purpose Agent Collaboration via Explainable Dynamic Decision Paths

Authors: Hanqing Yang, Hyungwoo Lee, Yuhang Yao, Zhiwei Liu, Kay Liu, Jingdi Chen, Carlee Joe-Wong

Organizations: CMU · Zoom · Salesforce · Amazon ·

Abstract

The increasingly popular agentic AI paradigm promises to harness the power of multiple, general-purpose large language model (LLM) agents to collaboratively complete complex tasks. While many agentic AI systems reduce complexity through predefined workflows or fixed agent roles, the ideal is to support truly autonomous agents capable of emergent collaboration across many interacting agents. Yet in practice, such unstructured interactions often lead to redundant work and cascading failures that are difficult to interpret or correct. In this work, we study multi-agent systems composed of general-purpose LLM agents that solve problems through emergent collaboration, without relying on predefined roles, control flows, or communication constraints. We introduce the Dynamic Interaction Graph (DIG), which captures emergent collaboration as a time-evolving causal network of agent activations and interactions. DIG makes emergent collaboration observable and explainable for the first time, enabling real-time identification, explanation, and correction of collaboration-induced error patterns directly from agents' collaboration paths. Thus, DIG fills a critical gap in understanding how general LLM agents solve problems together in truly agentic multi-agent systems. The project webpage can be found at: https://happyeureka.github.io/dig.

Figures & tables

Figure 3: Structural failure patterns in DIG, showing collaboration-invariant violations as interaction structure.

Figure 4: Task structures. Left: Count Frequency (fully parallelizable), where tasks are decomposed into independent subtasks that can be processed and merged once completed. Right: Research Job (dependency-constrained), where tasks are organized into stages with inter-stage dependencies, and violations cause processing or merging failures.

Setting	Agents	Load	# Failure ( $\downarrow$ )	Error ( $\downarrow$ )	Time ( $\downarrow$ )
Homogeneous (1000)
MAS-Only	4	4000	$\mathbf{2.11\pm 0.31}$	$0.083\pm 0.118$	$13.0\pm 9.1$
MAS+DIG	4	4000	$2.60\pm 1.80$	$\mathbf{0.025\pm 0.075}$	$\mathbf{10.7\pm 7.8}$
$\Delta$ %			$+23\%$	$-70\%$	$-18\%$
MAS-Only	4	8000	$\mathbf{2.60\pm 0.66}$	$0.688\pm 0.084$	$\mathbf{14.3\pm 8.4}$
MAS+DIG	4	8000	$9.10\pm 1.45$	$\mathbf{0.588\pm 0.138}$	$20.1\pm 9.4$

Table 1: CF with parallelizable tasks. Mean

\pm

std (10 runs). Bold = better.

\Delta\%

= MAS+DIG vs. MAS-Only.

Panel of Figure 5 — Figure 5: With DIG (a), the timeline of agent activations, event propagation, and edge-level rewrites indicates stable coordination and successful task termination. Using an LLM as judge (b), the DIG trace reveals unstable coordination, excessive waiting and rerouting, and delayed interventions.

Homogeneous (1000)
Setting	# Failure ( $\downarrow$ )	Error ( $\downarrow$ )	Time ( $\downarrow$ )
MAS-Only	$\mathbf{6.20\pm 2.96}$	$0.50\pm 0.30$	$24.8\pm 31.8$
MAS+DIG	$8.40\pm 1.69$	$\mathbf{0.28\pm 0.28}$	$\mathbf{17.9\pm 2.2}$
$\Delta$ %	$+35\%$	$-44\%$	$-28\%$
Heterogeneous ([…, 1000,100,10,1])
MAS-Only	$\mathbf{3.50\pm 1.43}$	$0.55\pm 0.15$	$\mathbf{43.4\pm 32.9}$

Table 2: RJ with two problem sets, each containing two sequential stages. Stage load is 4000 in the homogeneous setting and 1111 in the heterogeneous setting; each experiment uses 4 agents. Mean

\pm

std (10 runs). Bold = better.

\Delta\%

= MAS+DIG vs. MAS-Only.

Figure 6: MAS + DIG (20 agents): Large-scale DIG showing structured task propagation, controlled rerouting, and stable convergence despite dense interactions and high concurrency.

Appendix figures & tables5 assets

Supplementary material from the paper’s appendix.

Appendix

Domain	Setting	Detected Errors	Task Error	Runtime	MANOVA
		( $p$ )	( $p$ )	( $p$ )	( $p$ )
CF 4-agent	homo 4000	0.531	0.124	0.988	0.069
CF 4-agent	homo 8000	0.000 ∗	0.037 ∗	0.001 ∗	0.000 ∗
CF 4-agent	hetero 1111	0.178	0.864	0.715	0.712
CF 4-agent	hetero 2222	0.000 ∗	0.038 ∗	0.007 ∗	0.000 ∗
CF 6-agent	homo 6000	0.500	0.271	0.121	0.488

Appendix

Table 3: Significance test results comparing runs with and without DIG. Values report one-sided Mann–Whitney U test

p

-values for individual metrics and MANOVA

p

-values for the joint effect across all three metrics. Asterisks indicate significance at

p<0.05

Figure 7: Count Frequency example (Homogeneous_4000, 4 agents). The problem is within the group’s capability. Agent 1 decomposes the task for parallel execution, but due to asynchrony and partial observability, agents may ignore subtasks, reroute unnecessarily, or duplicate work. These stochastic behaviors make cooperation brittle, which DIG captures and helps diagnose.

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 1,039 papers

DHCG: Dynamic Construction of Hierarchical Collaboration Graphs for LLM-Based Multi-Agent Reasoning
Oct 6, 2026Jie Ren, Jiakang Yuan, Chenyu Huang +3Multi-Agent Collaboration Multi-Agent Reasoning
College of Future Information Technology, Fudan University, Shanghai, China · Shanghai Innovation Institute, Shanghai, China
What Do Agents Communicate? Characterizing Information Exchange in Multi-Agent Systems
May 19, 2026Yong Jin Chun, Iftekhar AhmedMulti-Agent Collaboration Multi-Agent Systems
Department of Informatics University of California, Irvine Irvine, CA, 92618
AgentWorld: Benchmarking Long-Horizon Collaboration of Multi-agent LLMs
Sep 25, 2026Raphael Shu, Yusen Zhang, Young Min Cho +7Multi-Agent Collaboration AI Agent Benchmarks
OpenAgents · Columbia University · University of Pennsylvania +2

Oct 6, 2026J/K move · Enter open · S save

DHCG: Dynamic Construction of Hierarchical Collaboration Graphs for LLM-Based Multi-Agent Reasoning

Jie Ren, Jiakang Yuan, Chenyu Huang, Hezeer Ma+2

College of Future Information Technology, Fudan University, Shanghai, China · Shanghai Innovation Institute, Shanghai, China

Multi-Agent Collaboration Multi-Agent Reasoning Multi-Agent Orchestration Multi-Agent Planning

Open paper AskPDF arXiv

LLM-based multi-agent systems (MAS) have demonstrated strong capabilities in solving complex problems across diverse domains. Recently, the dynamic orchestration of agent systems has become an important research direction. However, existing methods suffer from limited composition, misaligned dependencies, and inflexible scale, restricting their ability to adapt to reasoning requirements during execution. To address these limitations, we reframe MAS design as a partially observable Markov decision process, in which both the composition and scale of the MAS are dynamically determined. We propose DHCG, a novel framework that coordinates three modules (Planner, Worker, and Generator) to progressively construct a dynamic hierarchical collaboration graph from scratch based on the query and evolving execution feedback. At each step, guided by feedback, the Planner generates a set of distinct and complementary roles tailored to the current reasoning needs and selectively routes relevant information to each role. It can also finalize the hierarchical collaboration graph early or progressively expand it when additional reasoning is required. We further introduce action-aware preference optimization to train the Planner to make more effective decisions when constructing hierarchical collaboration graphs. We systematically evaluate DHCG across code generation, mathematical reasoning, and domain-specific reasoning benchmarks. DHCG achieves state-of-the-art average performance among the compared methods, improving over the single-agent baseline by 13.06 points and outperforming both static and dynamic MAS baselines by 2.77-8.02 points. Additional experiments further demonstrate its generalization across different Planner backbones and unseen Worker models.

Abstract

Multi-Agent LLM Systems

Explore similar work

DHCG: Dynamic Construction of Hierarchical Collaboration Graphs for LLM-Based Multi-Agent Reasoning

What Do Agents Communicate? Characterizing Information Exchange in Multi-Agent Systems

AgentWorld: Benchmarking Long-Horizon Collaboration of Multi-agent LLMs