AI agent automates software packaging for reliable builds

Vibenix: Agentic Software Packaging and Maintenance with Nix

Software Engineering

Summary

Software often depends on many hidden parts which makes building it difficult and unreliable. The authors built Vibenix, an AI helper that writes recipes for a tool called Nix to package software correctly. Vibenix uses an AI model to try building a package, learn from errors, and improve its instructions step by step. It successfully created working packages in over 95% of tests and fixed existing packages with high accuracy, making software building more trustworthy and easier for developers.

What this means in practice

  • For software developers: Automatically generate precise build instructions for complex software packages to save time and reduce errors.
  • For devops teams: Maintain and repair software packaging recipes automatically to ensure reliable and secure software supply.

Authors

Martim Monis, Martin Schwaighofer, Nuno Saavedra, Alexandra Mendes, Rene Mayrhofer, João F. Ferreira

Abstract

Implicit dependencies, uncontrolled network access, and other build-environment impurities create intransparent software supply chains. Identifying every dependency affecting a build is a difficult task. Functional package management addresses this through reproducible packaging, ensuring software relies exclusively on declared, hermetically isolated inputs. However, creating build recipes remains complex and time-consuming, posing a significant barrier to adoption for developers and maintainers. We present Vibenix, an open-source AI-powered agent that bridges this gap by automatically generating Nix packaging expressions. Vibenix uses a large language model (LLM) to iteratively refine recipes through a rule-guided build-repair feedback loop. On 466 packaging tasks from our new NixBench dataset, Vibenix produced buildable packages in 95.3% of cases using Claude Haiku 4.5, compared with 5.6% for a state-of-the-art LLM-based single-shot packaging tool. Manual validation of 100 packages revealed that 47% of the buildable packages were initially functionally correct; a post-build refinement process based on the evaluator-optimizer pattern increased functional correctness by 35 percentage points. On our new NixBench-Repair dataset of package maintenance tasks, Vibenix achieved an 87.6% success rate. These results demonstrate how unreliable LLMs can generate rigorously defined, complete, auditable dependency trees for real-world software, reducing maintainer burden while providing a trustworthy foundation for secure software distribution.