arXiv:2607. 02436v1 Announce Type: cross Abstract: Agentic coding assistants are increasingly given extra capabilities, such as browser based testing tools and design oriented system prompts, on the assumption that more capability yields better software.
The paper introduces rebuild‑dossier, an open‑source tool that locks an application’s real interface before code is written and enforces one‑test‑at‑a‑time building through automated checks. In experiments, a compliant agent failed a held‑back test while a rule‑breaking agent passed, showing that a passing test suite can be gamed. The study also demonstrates that the automated check mechanism, rather than interface‑locking alone, is crucial for reliable rebuilds, and that multi‑level verification catches errors that single‑level checks miss.
arXiv:2607. 11098v1 Announce Type: cross Abstract: Tool-using LLM agents are mostly evaluated assuming all tools work.
arXiv:2609.37315v1 Announce Type: cross Abstract: Tool-using agents are entering settings where a wrong action carries real cost, and the benchmarks certifying them grade what each simulated tool cal...
The paper introduces Traverse, a benchmark of 2,518 agent trajectories and 6,967 annotated mistakes across software engineering, computer use, and science tasks, revealing that failures often go unrecovered and can cause irreversible harm before a run is deemed successful. It shows that human judges struggle to detect the first mistake in most runs, while a 4‑billion‑parameter verifier called Scout can locate failures more effectively and improve task success when used to select among candidate runs. The study demonstrates that making failure detection inexpensive and reliable can enable long‑horizon agents to learn from their own mistakes and increase trustworthiness in autonomous AI.
arXiv:2605. 11030v2 Announce Type: replace-cross Abstract: Closed-loop tool-using agents are increasingly evaluated in executable web, code, and micro-task environments, but benchmark reports often conflate workloads, action-generating drivers, and the evidence admitted for systems-facing claims.
The paper introduces VerifierBench, a benchmark comprising 222 tasks from six agent benchmarks and 4,993 hidden test cases designed to evaluate the generation of executable verifiers by large language model agents. It presents Verifier4Verifier (V4V), a six-check audit that screens source verifiers and test suites, and reports that across 19 models, normalized balanced accuracy ranges from 10.6 to 62.1, with generated verifiers more reliably rejecting invalid solutions than accepting valid ones. Case studies highlight that many false rejections stem from verifiers that accept no tested solution, revealing unsupported assumptions about solution formats, tool use, and code changes that exclude valid solutions, and underscore the need to test both acceptance of correct solutions and other permissible solutions.
The paper investigates the reliability of software produced by agentic AI by comparing AI-generated versions of ten well-known Linux utilities to their human-written counterparts. Using fuzz testing (both black-box and coverage-guided AFL++), the authors find that AI-generated code is often as reliable or more reliable than the latest human versions, with fewer memory errors but a higher incidence of hangs. The study emphasizes that robust AI-generated software requires careful prompting, skilled human oversight, and that the AI workflow can serve as a cost-effective specification for sustainable code.
AgentSWE is a new benchmark for agent software engineering that evaluates coding agents across three lifecycle stages: creating an agent from a natural‑language requirement, editing existing agent codebases, and optimizing toward a measurable target. The benchmark consists of 25 commission‑style tasks, where builders iterate on a few illustrative cases and are then tested on held‑out cases that cover the full requirement, with hard pass/fail gates or objective‑based grading. Experiments show that current coding agents fail to meet the requirements, revealing common failure modes such as overfitting, incomplete runtimes, and unsupported completion claims.
arXiv:2607. 04542v1 Announce Type: cross Abstract: Every LLM agent run re-derives its behavior token by token on a frontier model: brilliant, expensive, slow, and unbounded.
arXiv:2610.17895v2 Announce Type: cross Abstract: Coding agents are tasked with producing sophisticated interactive applications such as websites, desktop applications, and games. While humans assess...
WatchPoint is a simulated‑user system that generates and runs diagnostic scripts against a live web application, producing structured observations to guide coding model retries. Unlike prior methods that rely on screenshots or non‑executable metrics, WatchPoint operates on Web‑Bench—a benchmark of 50 multi‑file web projects with 1,000 sequential tasks verified by deterministic end‑to‑end tests. It recovers 57.6% of diagnosed tasks, matching a human tester’s 54.5% recovery rate, and identifies when such simulated feedback is beneficial or should be withheld.