Loading article...
Loading article...
Stream/Area:
-
Subject:
-
Paper ID:
IJSREM68739
Volume:
10 |
11 |
Nov 2026
Pages:
1-11
Publication Frequency:
Monthly (12 issues per year)
License:International Journal of Scientific Research in Engineering and Management follows the Creative Commons Attribution 4.0 (CC BY 4.0) License, allowing free use and distribution with proper attribution.
Computer Engineering
SVPM`s College of Engineering
Computer Engineering
SVPM`s College of Engineering
Computer Engineering
SVPM`s College of Engineering
Computer Engineering
SVPM`s College of Engineering
Computer Engineering
SVPM`s College of Engineering
Web applications change so often that manual testing cannot keep up, and scripted automation breaks whenever the user interface changes. Large language models (LLMs) and the agent systems built on them offer a different approach: software agents that observe an application, decide what to do, act through a browser, and judge the result. This paper reviews how such agents are applied to web application testing. We organize existing work around five stages of the testing lifecycle (requirement analysis, test generation, execution, oracle and defect detection, and reporting), compare single-agent and multi-agent architectures, and examine how agentic testers are evaluated. We collect the challenges that keep recurring: hallucination, non-determinism, cost, weak evaluation practice, and safety. The reviewed work automates isolated stages, and a system that handles the whole lifecycle with little manual scripting has not yet been demonstrated. We therefore propose a five-agent workflow, built on LangGraph, Playwright, a locally hosted open-source model, and PostgreSQL, for end-to-end autonomous testing, and we outline a protocol for evaluating it. The paper presents a design and a literature-based analysis only; it reports no experimental results of its own.
KEYWORDS
Agentic AI turns web testing from the replay of fixed scripts into goal-driven interaction with an application. The literature shows that language models help generate tests, that agents can execute natural-language test cases and explore applications for usability problems, and that a lifecycle view is a productive way to organize the field. It also shows that the results are preliminary, that evaluation is inconsistent, and that no existing work convincingly covers the whole lifecycle. Hallucination, non-determinism, cost, oracle reliability, and safety must be addressed before such systems can be trusted in production. In our view, multi-agent pipelines with explicit orchestration, shared benchmarks, and human checkpoints are the most promising route. This work has limitations. The review is narrative and not exhaustive, many of the cited findings come from preprints, and the proposed platform has been designed but not yet implemented or evaluated. Our next step is to build the prototype and to run the evaluation protocol of Section III-D, so that the claims made here can be tested against measured results.
[1] A. Chevrot, A. Vernotte, J.-R. Falleri, X. Blanc, and B. Legeard, “Are autonomous web agents good testers?,” Proc. ACM Softw. Eng., vol. 2, no. ISSTA, pp. 206–228, 2025. [2] M. Hammoudi, G. Rothermel, and P. Tonella, “Why do record/replay tests of web applications break?,” in Proc. IEEE Int. Conf. Software Testing, Verification and Validation (ICST), 2016, pp. 180–190. [3] M. Leotta, B. García, F. Ricca, and J. Whitehead, “Challenges of end-to-end testing with Selenium WebDriver and how to face them: A survey,” in Proc. IEEE Conf. Software Testing, Verification and Validation (ICST), 2023, pp. 339–350. [4] L. Wang et al., “A survey on large language model based autonomous agents,” Front. Comput. Sci., vol. 18, no. 6, p. 186345, Dec. 2024. [5] G. J. Myers, C. Sandler, and T. Badgett, The Art of Software Testing, 3rd ed. Hoboken, NJ, USA: Wiley, 2012. [6] J. Wang et al., “Software testing with large language models: Survey, landscape, and vision,” IEEE Trans. Softw. Eng., vol. 50, no. 4, pp. 911–936, Apr. 2024. [7] R. Santos, I. Santos, C. Magalhaes, and R. de Souza Santos, “Are we testing or being tested? Exploring the practical applications of large language models in software testing,” in Proc. IEEE Conf. Software Testing, Verification and Validation (ICST), 2024, pp. 353–360. [8] S. Yao et al., “ReAct: Synergizing reasoning and acting in language models,” in Proc. Int. Conf. Learning Representations (ICLR), 2023. [9] P. P. Kumaresen, M. Frasheri, and E. P. Enoiu, “Agent-based software testing: A definition and systematic mapping study,” in Proc. IEEE 20th Int. Conf. Software Quality, Reliability and Security Companion (QRS-C), 2020, pp. 24–31. [10] S. Paydar and M. Kahani, “An agent-based framework for automated testing of web-based systems,” J. Softw. Eng. Appl., vol. 4, pp. 86–94, 2011.
Publication Date
11/5/2026
Peer Review Type
Double-Blind Peer Review
Review Rounds: 2
Documents
Article Views
1
Total views since publication
PDF Downloads
6
Total downloads since publication
Geographic Reach
5
Digital Preservation: This article is permanently archived on Zenodo and complies with all data preservation standards across major indexing platforms.
Plagiarism Screening
This manuscript has been checked for plagiarism using TurnitinPlagiarismChecker software.
Conflict of Interest
The authors declare no conflict of interest related to this research work.
Funding Information
No funding information available for this research.
Related research in the same subject area
Loading similar articles...