Prototype, artefacts and benchmarks for paper: Realisability-Driven Repair of Natural-Language Requirements via Neuro-Symbolic Reasoning

Prototype, artifacts, and benchmarks for the paper “Realisability-Driven Repair of Natural-Language Requirements via Neuro-Symbolic Reasoning.” This artifact accompanies the paper on SpecRepair, a Docker-packaged tool for constructing and repairing realizable specifications from natural-language requirements. It contains the exported Docker image, configuration template, usage instructions, benchmark datasets, and benchmark documentation. The archive supports reproducing the paper example and experiments and running SpecRepair on custom requirement files using a user-provided LLM configuration.