HGM-Dual: Lineage-Aware Guided Agent Harness Optimization for Long-Horizon Planning

Sudipta Paul; Pengrui Quan; Aounon Kumar; Nicholas Tzou; Vivek Kulkarni; Srinivas Chappidi

cs.AIOct 21, 2026

HGM-Dual: Lineage-Aware Guided Agent Harness Optimization for Long-Horizon Planning

Authors: Sudipta Paul, Pengrui Quan, Aounon Kumar, Nicholas Tzou, Vivek Kulkarni, Srinivas Chappidi

Organizations: Samsung Research America

Abstract

Harness optimization, which improves an agent's prompts, tools, and workflows, is an increasingly effective way to enhance agents built on frozen or closed-source LLMs. Leading approaches such as the Darwin--Godel Machine and the Huxley--Godel Machine (HGM) rely on recursive self-improvement and were developed mainly for coding, where the competence needed to improve the agent coincides with the competence being improved. Planning tasks, in which an agent must produce a sequence of tool calls that jointly satisfies interacting constraints, break this alignment. There, open-ended self-modification repeatedly explores ineffective changes and consumes substantial evaluation budgets. We introduce HGM-Dual, a diagnostic-guided framework for budget-efficient harness evolution for planning agents. First, HGM-Dual represents agent evolution as lineage-aware context. This context records what each ancestor changed, which behaviors improved or regressed, and which modifications were ineffective, giving the editor explicit edit-outcome attribution instead of an unstructured archive of past agents. Second, HGM-Dual introduces error-conditioned dual-path child generation. For each selected parent, it generates a general child that preserves open-ended exploration and specialized children that target the harness components most likely responsible for the parent's highest-ranked constraint violations. HGM-Dual outperforms the strongest baseline by up to 14.45% relative improvement in case accuracy while matching HGM variants' final performance with 2x-4x fewer evaluation budget.

Figures & tables

Figure 1: Edit behavior in HGM. Among child nodes, 41.8% repeat previous edits, 19.4% target an error category absent from the parent’s observed failures, and 64.2% retain the parent’s dominant error category. See Appendix A.1 for details.

Figure 2: HGM search: select a parent, generate an open-ended edit, evaluate the child, and add it to the archive.

Figure 3: Overview of HGM-Dual and one expansion round. Given a selected parent

\mathcal{H}_{i}

, HGM-Dual constructs an evolution context

\mathcal{Z}_{i}

by summarizing edits and outcomes along its ancestral path

\mathrm{Anc}(\mathcal{H}_{i})

. The evolution context is provided to both editing paths: a general editor proposes an open-ended improvement, while specialized editors target the highest-ranked errors in the parent’s error profile

\varepsilon_{i}

. The resulting children are evaluated under the same protocol, and only the strongest child is added to the archive.

Panel of Figure 4 — Algorithm 1 HGM-Dual Expansion

		Travel		Shopping
Method	Evaluation Budget	Composite Score	Case Acc. (%)	Composite Score	Case Acc. (%)
Seed Agent	0	76.46 $\pm$ 0.22	25.00	81.91 $\pm$ 1.17	50.83
HGM	4000	77.06 $\pm$ 0.53	30.42	83.12 $\pm$ 0.27	50.83
HGM-CA	4000	83.23 $\pm$ 2.65	54.17	85.32 $\pm$ 0.90	58.33
HGM-CF	4000	84.11 $\pm$ 0.07	56.25	86.67 $\pm$ 1.99	60.83
HGM-Dual	4000	87.77 $\pm$ 1.31	64.38	89.23 $\pm$ 1.33	66.67

Table 1: Performance comparison of task agent optimization for Travel and Shopping at an evaluation budget of 4000. Best per column in bold , second best underlined .

Figure 4: Budget efficiency and error breakdown. (a, b) Best agent performance on {1k, 2k, 3k, 4k} evaluation budgets for travel and shopping task. (c, d) Per-category change in failure rate of the best agent after optimization relative to the seed agent on the

60

held-out task instances. Right-hand column gives the seed agent’s failure rate, filled markers denote

q<0.05

		Travel		Shopping
Method	Evaluation Budget	Composite Score	Case Acc.	Composite Score	Case Acc.
Seed Agent	0	76.46 $\pm$ 0.22	25.00	81.91 $\pm$ 1.17	50.83
ADAS	1000	80.83 $\pm$ 2.80	35.00	81.98 $\pm$ 2.15	51.25
GEPA	1000	81.67 $\pm$ 0.66	39.58	81.18 $\pm$ 0.96	42.92
HGM	1000	77.06 $\pm$ 0.53	30.42	83.12 $\pm$ 0.27	50.83
HGM-CA	1000	80.65 $\pm$ 0.26	41.67	84.46 $\pm$ 0.38	53.75

Table 2: Performance comparison for Travel and Shopping at an evaluation budget of 1000.

Table 8

Appendix figures & tables1 asset

Supplementary material from the paper’s appendix.

Appendix

	Unoptimized	Optimized
Travel	$400–800	$520–1,010
Shopping	$140–360	$300–910

Appendix

Table 6: Estimated task-agent API cost (USD) of 1000 samples evaluations on Travel and Shopping. The task agent is GPT-5.4-mini (high reasoning effort). Optimized averaged candidate agents evaluated by HGM and HGM-Dual. Lower bounds assume

{\sim}80\%

of input tokens are served from the prompt cache; upper bounds assume no caching.

Agent Harness Optimization

Agent harness optimization improves the prompts, tools, memory, workflows, and control code that shape how AI agents act.

See all 185 papers

Aug 3, 2026

HarnessCompass: Guiding Automatic Harness Evolution toward Generalizable and Effective Agent Harnesses

Harness design plays a critical role in agent performance by shaping how large language models (LLMs) perceive, reason over, and act within executable environments. Recent work has proposed automatic harness evolution, which iteratively improves the harness from agent--environment interactions. However, existing methods often overfit to the evolution tasks, rely exclusively on trajectory-derived signals, and optimize harness components jointly, causing interference across components. We propose HarnessCompass, a novel automatic harness evolution framework built around constrained evolution, proactive feedback, and component-wise optimization. HarnessCompass first enforces global constraints on evolution, restricting modifications to task-agnostic harness changes that generalize beyond the evolution tasks. It then augments trajectory-derived evidence with proactive first-person feedback from the agent about harness usage, yielding richer signals for evolution. Finally, it decouples the optimization of different harness components before consolidating them into a unified harness, reducing cross-component interference while preserving component synergy. On SWE-bench Verified with GPT-5.4, HarnessCompass improves Pass@1 from 54% to 66% in only 5 evolution iterations, outperforming AHE in both effectiveness and evolution efficiency. In addition, the evolved harness transfers effectively to held-out tasks and other models, demonstrating substantially stronger generalization than prior automatic harness evolution methods.

Luan Zhang, Ruochen Zhou, Dandan Song +9

Beijing Institute of Technology, China · City University of Hong Kong, China · Independent, China

Agent Harness Optimization LLM Agents LLM Agent Harnesses

Ask PDF

Jun 12, 2026

HarnessX: A Composable, Adaptive, and Evolvable Agent Harness Foundry

AI agent performance depends critically on the runtime harness, comprising the prompts, tools, memory, and control flow that mediate how a model observes, reasons, and acts. Yet today's harnesses remain largely hand-crafted and static: each new model or task still demands bespoke scaffolding, and the rich traces produced during execution are rarely distilled back into systematic improvement. We introduce HarnessX, a foundry for composable, adaptive, and evolvable agent harnesses. HarnessX assembles typed harness primitives via a substitution algebra, adapts them through AEGIS, a trace-driven multi-agent evolution engine grounded in an operational mirror between symbolic adaptation and reinforcement learning, and closes the harness-model loop by turning trajectories into both harness updates and model training signal. Across five benchmarks (ALFWorld, GAIA, WebShop, tau^3-Bench, and SWE-bench Verified), HarnessX yields an average gain of +14.5% (up to +44.0%), with gains largest where baselines are lowest. These results suggest that agent progress need not come from model scaling alone: composing and evolving runtime interfaces from execution feedback is an actionable and complementary lever. The complete codebase will be open-sourced in a future release.

Tingyang Chen, Shuo Lu, Kang Zhao +11

Darwin Agent Team

LLM Agent Orchestration Agent Harness Optimization LLM Agent Workflow Optimization LLM Agent Self-Improvement LLM Agent Harnesses Self-Evolving Agents

Ask PDF

Oct 10, 2026

Harmony Harness: Multi-Objective Self-Evolution of Agent Harnesses

Recent work on self-evolving LLM agents treats the agent harness as an optimization target that can be iteratively improved through execution feedback. Existing approaches, however, largely optimize a single objective and evolve the harness as a monolithic artifact, limiting their ability to handle competing goals and to separate transferable design from environment-specific adaptation. We introduce Harmony Harness, a framework for multi-objective, compositional harness evolution. Harmony Harness formulates harness evolution as a multi-objective optimization problem and uses decomposition-based evolutionary search to maintain solutions across different regions of the Pareto frontier, avoiding reliance on a hand-designed structural diversity metric for repository-scale harnesses. In parallel, we introduce factorized evolution operators that separately refine general-purpose harness mechanisms and benchmark-specific adaptations, together with a dynamic composition workflow that integrates them at inference time. Across SecRepoBench and BaxBench, Harmony Harness consistently outperforms strong harness-evolution baselines, with gains exceeding 10% on both benchmarks. Our results show that explicitly modeling both objective trade-offs and harness structure yields substantially stronger and more transferable agent evolution than existing single-objective, monolithic approaches.

Weichen Yu, Lesly Miculicich, Bhavana Dalvi Mishra +5

Google Cloud AI Research · Carnegie Mellon University · Google DeepMind

Agent Harness Optimization LLM Agent Harnesses Self-Evolving Agents Multi-Objective Evolutionary Optimization

Ask PDF

Abstract

Agent Harness Optimization

Explore similar work

HarnessCompass: Guiding Automatic Harness Evolution toward Generalizable and Effective Agent Harnesses

HarnessX: A Composable, Adaptive, and Evolvable Agent Harness Foundry

Harmony Harness: Multi-Objective Self-Evolution of Agent Harnesses