Skip to content
#data science #software testing Review Open access

Replication package for "Production-Oriented Evaluation and Testing of LLM-Based Agents: A Systematic Literature Review"

Oct 2026 · Zenodo (CERN European Organization for Nuclear Research)
Software Engineering Techniques and Practices Software Engineering Research

Abstract

Replication package for the systematic literature review "Production-Oriented Evaluation and Testing of LLM-Based Agents: A Systematic Literature Review" (revised version, submitted to Information and Software Technology; first submitted as "Evaluation and Testing of LLM-Based Agents in Production: A Systematic Literature Review"). The review follows Kitchenham–Charters guidelines with PRISMA 2020 reporting. The main search covers Scopus, Web of Science, IEEE Xplore and the ACM Digital Library up to 24 March 2026; with citation tracking it yields a review corpus of 88 primary studies (74 from the main search, 14 from citation tracking). A search extension to 25 September 2026 with the same query and criteria yields 314 further studies, reported as a separate update analysis. A Scopus sensitivity search with broader production-oriented terms (300-record random sample) estimates the recall of the query. Version 2.0.0 accompanies the revised article. Version 1.0.0 (DOI 10.5281/zenodo.20583928) accompanies the article as first submitted and is preserved unchanged in the folder 2026-03/. Contents: corpus/, extraction/, risk-of-bias/ — the review corpus of 88: included and excluded studies with screening decisions, dual data extraction (Reviewer 1, Reviewer 2 and arbiter-resolved; 27 analytical fields) and three-domain quality assessment. 2026-03/ — the March 2026 round exactly as released in v1.0.0 (80 studies, agreement reports, audits). 2026-09/ — the September 2026 round: ACM Digital Library and update searches (full-text records R001–R670), Scopus sensitivity search, extraction and quality assessment of 335 studies, agreement reports, and the single-label recode of automated-testing with its held-out validation. primary-studies.bib (88) and 2026-09/primary-studies.bib (335). reproduce.ipynb — a self-contained notebook with 595 automated checks that reproduce every statistic of the article computed from the per-record data (Section 2: review corpus; Section 3: September round; Appendix A: the 2026-03 round as first submitted). README.md (structure, data dictionary, errata), CHANGELOG.md, methodology.md, statistical-analyses.md, prisma-2020-checklist.md, kitchenham-2007-compliance.md, requirements.txt, LICENSE. To reproduce: install the dependencies (pip install -r requirements.txt) and run reproduce.ipynb top to bottom. Expected output: 595 checks passed, 0 failed. Licensed under CC BY 4.0.

View source

Similar papers

#computer vision Review Sep 2017

Agile Software Development Methods: Review and Analysis

This publication proposes a definition and a classification of agile software development approaches and analyses ten software development methods that can be characterized as being "agile" against the defined criterion.

P. Abrahamsson, O. Salo, Jussi Ronkainen et al. · 727 citations · ⚡54
#computer vision Jun 2008

The impact of agile practices on communication in software development

The study shows that agile practices improve both informal and formal communication, but indicates that, in larger development situations involving multiple external stakeholders, a mismatch of adequate communication mechanisms can sometimes even hinder the communication.

M. Pikkarainen, Jukka Haikara, O. Salo et al. · 401 citations · ⚡48
#machine learning Review Open access Oct 2014

Software development in startup companies: A systematic mapping study

The results indicate that software engineering work practices are chosen opportunistically, adapted and configured to provide value under the constrains imposed by the startup context.

Nicolò Paternoster, Carmine Giardino, M. Unterkalmsteiner et al. · 394 citations · ⚡54

Trajectory Balance: Improved Credit Assignment in GFlowNets

It is proved that any global minimizer of the trajectory balance objective can define a policy that samples exactly from the target distribution, and empirically demonstrate the benefits of the trajectories balance objective for GFlowNet convergence, diversity of generated samples, and robustness to long action sequenc...

Esmeralda S. Whitammer, Moksh Jain, Emmanuel Bengio et al. · 302 citations · ⚡60

Related blog posts

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.