ai4s-research/ai4s-skills

AI4S Agent runs an end-to-end AI4S research pipeline from a broad direction or topic into literature survey, experiment package, and paper. It is a meta-skill that orchestrates four downstream markdown skills in order.

Experiment Suite builds a complete research experiment package with design docs, runnable code, results, publication-grade figures, and a structured report. Use it when a research question needs an end-to-end experiment workflow with honest provenance.

Integrity Auditor audits papers for image misuse, numerical anomalies, and logical gaps, then writes a reviewable evidence report. Use it for PDFs, DOI or arXiv papers, or outputs from a local paper-writer run.

Literature Survey builds a full research survey on a specific topic, with LaTeX source, classified citations, figures, and a publication table. Use it when you need a structured literature review for a thesis chapter, grant, or field overview.

Mindmap Render generates high-resolution mindmaps from Markdown unordered lists or structured outlines. It can export interactive HTML, PNG, and PDF for presentations or review.

Paper Writer builds a publication-style research paper on a specific topic, including citations, figures, sections, and PDF output. It is used when an agent needs to produce a complete paper from scratch or format provided results into a manuscript.

Research Explorer helps turn a broad research direction into candidate topics with feasibility and innovation scoring. It also gathers a 20–30 paper pre-survey when the user needs a research landscape overview.