Xu Shuyuan

You can plan papers from topic intake to blueprint, search Semantic Scholar for citations, engineer reproducible experiments with config-driven training, analyze ablation tables into conservative claims, draft and polish paper sections, design publication figures, and run final pre-submission and defense audits.

Graduate students, thesis and dissertation writers, and academic researchers managing the full literature-to-experiment-to-writing pipeline. They suit users with advisor-given topics who need research blueprints, experiment-to-claim maps, citation plans, and defense-ready submission checks.

The experiment-engineering and autoresearch-loop skills provide project skeletons, config-driven training, smoke tests, output manifests, resume state, and verify/guard gates. They coordinate local_mac smoke tests with remote_desktop_4060 formal GPU runs and map code runs to thesis experiments.

Skills are invoked by description-matched triggers within a skill-capable environment. Some reference specific setups such as local_mac smoke testing and a remote_desktop_4060 GPU machine for formal runs, plus standard research artifacts like PDFs, DOCX files, BibTeX, and Mermaid diagrams.

The manifest does not specify licensing or cost terms. The author states a goal of giving back to the community, and the skills are published publicly in the registry, but users should check the repository for explicit license details before redistribution.