Paper page - DarwinX: Evolving Agent Harnesses Through Natural Selection

Source: https://huggingface.co/papers/2608.07545 Authors:

,

,

,

,

,

,

,

,

,

,

Abstract

DarwinX evolves agent harnesses via population selection with frozen models, improving verified performance across benchmarks without benchmark-specific patches.

AnLLM agent’s capability depends not only on model weights but on itsharness: prompts, tools, skills, and control flow.Self-improvementloops already editharnesses, yet single-lineage search is path-dependent and local wins often regress other tasks. We introduceDarwinX, which treats self-evolution as selection over a population ofharnesses with the model frozen: apreserve-and-extend contractadmits only variants that extend coverage without regressing, anarchivekeeps alternative lineages forrecombination, and failure-, teacher-, and self-derived evidence share one edit interface. Fitness comes from each benchmark’s ownverifier: no gold solutions, no hand-picked winners. Across four benchmarks that progressively separate the evolution signal from the test, one loop adds about 17 points on average: Terminal-Bench 2.1 rises +7.7 to 83.2% on a matched base and to the verified frontier at 84.7% on a stronger one; TerminalWorld’s held-out split reaches 68.3%, ahead of every off-the-shelf agent; WebArena-Infinity real-task pass@1 rises from 43.5% to 93.0% audit-clean; and a Terminal-Bench 2.1harnesstransfers unchanged to SWE-bench Verified. What evolves is general agent competence, not benchmark-specific patches, so it survives changes of task,verifier, and base model. A frozen model need not be a fixed agent:harnessselection turns evaluation compute into durable capability.

View arXiv pageView PDFProject pageAdd to collection

Get this paper in your agent:

hf papers read 2608\.07545

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2608.07545 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2608.07545 in a dataset README.md to link it from this page.

Spaces citing this paper1

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.