arXiv subjects

Emna Ksontini

Publications and source records attributed to Emna Ksontini.

7 recordsLinked to original sources

arXiv2026

A Large-Scale Dataset of MCP Implementations on GitHub

The rapid emergence of the Model Context Protocol (MCP) has introduced a new standard for connecting large language models to external tools and services. Despite its rapid adoption in open-source development, systematic understanding of how MCP is implemented, structured, and maintained remains limited. This study presents the first large-scale, evidence-based dataset of real-world MCP implementation collected directly from GitHub. Using a hybrid pipeline that integrates the GitHub REST and GraphQL APIs with custom Python verification scripts, 3,238 candidate repositories were discovered, filtered, and validated through multi-stage evidence checks. Each verified project was classified by operational role (e.g., client, server, gateway) and exported in a reproducible JSONL schema. A manual review of a representative subset confirmed an overall precision of 83% at a 95% confidence level, and additionally revealed a set of repositories functioning primarily as educational samples, tutorials, or demonstration templates. A targeted exclusion rule was then applied to remove these non-operational repositories, resulting in a final dataset of 2,297 validated MCP projects. The analysis shows that Python and TypeScript dominate MCP development, with hybrid architectures emerging as the most common design pattern. By emphasizing transparent verification strategies, structured evidence tagging, and reproducible data organization, this work establishes a foundational benchmark for studying real-world MCP ecosystems and supports future research on integration, connectivity, and compatibility across the broader developer community.

cs.SE↗

arXiv2026

ML in a Box: Analyzing Containerization Practices in Open Source ML Projects

Containerization has become increasingly essential in the machine learning (ML) domain, providing reproducibility, portability, and environment consistency. While prior studies have analyzed Dockerfile structures and best practices, none have examined ML projects in depth to reveal how the iterative nature of ML workflows influences container footprint, build performance, and caching behavior. We present the first large scale empirical study of 1,993 ML related Dockerfiles, combining quantitative analysis of container roles in ML projects and build dynamics with a qualitative investigation of refactoring practices. Results show that containers serve distinct roles across training, inference, and infrastructure. Containers are typically large, averaging 10.27 GB in size, and require long build times of about 8.84 minutes. We find that 44.4% of commits trigger rebuilds, primarily due to context file changes (96.4%), with experimentation being the main motive behind those commits that initiate rebuilds. Despite partial cache reuse, 71% of rebuild work is wasted on redundant computation. From stable projects, we identify 7 recurring ML-specific Dockerfile refactoring patterns that improve build efficiency and reduce container footprint.

cs.SE↗

arXiv2025

CARGO: A Framework for Confidence-Aware Routing of Large Language Models

As large language models (LLMs) proliferate in scale, specialization, and latency profiles, the challenge of routing user prompts to the most appropriate model has become increasingly critical for balancing performance and cost. We introduce CARGO (Category-Aware Routing with Gap-based Optimization), a lightweight, confidence-aware framework for dynamic LLM selection. CARGO employs a single embedding-based regressor trained on LLM-judged pairwise comparisons to predict model performance, with an optional binary classifier invoked when predictions are uncertain. This two-stage design enables precise, cost-aware routing without the need for human-annotated supervision. To capture domain-specific behavior, CARGO also supports category-specific regressors trained across five task groups: mathematics, coding, reasoning, summarization, and creative writing. Evaluated on four competitive LLMs (GPT-4o, Claude 3.5 Sonnet, DeepSeek V3, and Perplexity Sonar), CARGO achieves a top-1 routing accuracy of 76.4% and win rates ranging from 72% to 89% against individual experts. These results demonstrate that confidence-guided, lightweight routing can achieve expert-level performance with minimal overhead, offering a practical solution for real-world, multi-model LLM deployments.

cs.SE↗

arXiv2025

FaaSGuard: Secure CI/CD for Serverless Applications -- An OpenFaaS Case Study

Serverless computing significantly alters software development by abstracting infrastructure management and enabling rapid, modular, event-driven deployments. Despite its benefits, the distinct characteristics of serverless functions, such as ephemeral execution and fine-grained scalability, pose unique security challenges, particularly in open-source platforms like OpenFaaS. Existing approaches typically address isolated phases of the DevSecOps lifecycle, lacking an integrated and comprehensive security strategy. To bridge this gap, we propose FaaSGuard, a unified DevSecOps pipeline explicitly designed for open-source serverless environments. FaaSGuard systematically embeds lightweight, fail-closed security checks into every stage of the development lifecycle-planning, coding, building, deployment, and monitoring-effectively addressing threats such as injection attacks, hard-coded secrets, and resource exhaustion. We validate our approach empirically through a case study involving 20 real-world serverless functions from public GitHub repositories. Results indicate that FaaSGuard effectively detects and prevents critical vulnerabilities, demonstrating high precision (95%) and recall (91%) without significant disruption to established CI/CD practices.

cs.SE↗

arXiv2025

From REST to MCP: An Empirical Study of API Wrapping and Automated Server Generation for LLM Agents

The Model Context Protocol (MCP) is emerging as a standard interface through which LLM agents invoke external tools, and a growing ecosystem of MCP servers now mediates access to vendor services. Most of these servers target vendors that already expose REST APIs, yet the relationship between MCP tool interfaces and the underlying API surface has not been empirically characterised. This paper presents the first large-scale study of MCP server construction. We analyse 116 official servers to determine REST reliance and integration strategies (RQ1); examine servers paired with OpenAPI specifications to quantify operation exposure, omission, and mapping patterns (RQ2); evaluate automated generation from 80 real-world OpenAPI contracts (RQ3); and assess specification repair and tool-set transformations to improve correctness and reduce complexity (RQ4). We find that 88.6% of servers are fully or partially REST-backed, with 92% implementing tools as bare API wrappers. MCP servers expose a median of 19% of available operations, following systematic patterns predictable from the specification. Baseline generation succeeds for 76% of sampled tools; automated repair raises this to 94.2%, while filtering and regrouping reduce the median tool count per API by one-third. We release AutoMCP, an end-to-end pipeline integrating specification repair and empirically grounded tool-set transformations.

cs.SE↗

arXiv2025

Scalable and Cost-Efficient ML Inference: Parallel Batch Processing with Serverless Functions

As data-intensive applications grow, batch processing in limited-resource environments faces scalability and resource management challenges. Serverless computing offers a flexible alternative, enabling dynamic resource allocation and automatic scaling. This paper explores how serverless architectures can make large-scale ML inference tasks faster and cost-effective by decomposing monolithic processes into parallel functions. Through a case study on sentiment analysis using the DistilBERT model and the IMDb dataset, we demonstrate that serverless parallel processing can reduce execution time by over 95% compared to monolithic approaches, at the same cost.

cs.DC↗

arXiv2025

Refactoring for Dockerfile Quality: A Dive into Developer Practices and Automation Potential

Docker, the industry standard for packaging and deploying applications, leverages Infrastructure as Code (IaC) principles to facilitate the creation of images through Dockerfiles. However, maintaining Dockerfiles presents significant challenges. Refactoring, in particular, is often a manual and complex process. This paper explores the utility and practicality of automating Dockerfile refactoring using 600 Dockerfiles from 358 open-source projects. Our study reveals that Dockerfile image size and build duration tend to increase as projects evolve, with developers often postponing refactoring efforts until later stages in the development cycle. This trend motivates the automation of refactoring. To achieve this, we leverage In Context Learning (ICL) along with a score-based demonstration selection strategy. Our approach leads to an average reduction of 32% in image size and a 6% decrease in build duration, with improvements in understandability and maintainability observed in 77% and 91% of cases, respectively. Additionally, our analysis shows that automated refactoring reduces Dockerfile image size by 2x compared to manual refactoring and 10x compared to smell-fixing tools like PARFUM. This work establishes a foundation for automating Dockerfile refactoring, indicating that such automation could become a standard practice within CI/CD pipelines to enhance Dockerfile quality throughout every step of the software development lifecycle.

cs.SE↗