arXiv:2609.15387v3 Announce Type: replace-cross Abstract: Human evaluation provides a direct measure of the quality of LLM-generated web applications. However, fitting human judgments through automat...
arXiv:2609.15387v1 Announce Type: cross Abstract: Human evaluation provides a direct measure of the quality of LLM-generated web applications. However, fitting human judgments through automated evalu...
AgentHorizon is a benchmark comprising 1,373 computer‑use tasks with instruction‑trajectory pairs collected from 166 hours of human‑recorded activity across three operating systems. The benchmark uses a paired design that swaps instructions to create negative tasks, testing judges’ ability to distinguish successful trajectories from those that completed a similar but incompatible request. Eleven judges were evaluated, with GPT‑5.5 achieving the highest balanced accuracy of 80.9% on the full trajectory subset, while tool‑use benefits some models but harms others, revealing large variability in judges’ performance.
arXiv:2609.21293v1 Announce Type: new Abstract: Autonomous software generation (ASG) aims to turn human requirements into executable applications, but delivering these applications does not necessari...
arXiv:2606. 30573v1 Announce Type: new Abstract: We introduce SWE-Interact, a new testbed for evaluating coding agents on multi-turn, interactive, user-driven software engineering tasks.
arXiv:2608.28795v1 Announce Type: cross Abstract: Modern artificial-intelligence coding agents can be equipped with tools for checking their own work e.g. a linter, a boot probe, a shell, a screensho...
arXiv:2607. 20536v1 Announce Type: new Abstract: Tool-use agents that address day-to-day digital tasks such as ordering groceries must not only operate applications, but also interact with the user, e.
arXiv:2610.18760v1 Announce Type: new Abstract: Test-time scaling for large language models (LLMs) can involve generating multiple candidate solutions for selection or revising an earlier attempt. To...
The paper presents EvalAgent, an AI assistant that automates agent evaluation by encoding domain expertise into evaluation skills such as procedural instructions, reusable code, and dynamic API retrieval. EvalAgent constructs a trace-based pipeline that outputs metrics, executable code, and reports, and is evaluated using a new meta-evaluation framework and AgentEvalBench. Results show that EvalAgent improves the Eval@1 metric from 17.5% to 65% and receives 79.5% human expert preference, while ablation studies confirm the importance of evaluation skills.
arXiv:2609.22000v1 Announce Type: new Abstract: Computer-use agents (CUAs) have advanced along two separate lines: graphical interaction and software development through code and the command line. Re...
arXiv:2608.27831v1 Announce Type: cross Abstract: Coding agents are now commonly evaluated on the SWE-bench family of benchmarks, whose tasks are built from curated GitHub issues--long, structured, a...
arXiv:2510. 09801v3 Announce Type: replace Abstract: While benchmarks measure the accuracy of LLM-powered agents, they mostly assume full automation, failing to represent the collaborative nature of real-world use cases.