license: informal permission in README ("Use freely. Attribution appreciated but not required."); no licence file · 10 skills · last update: 2026-08-26
Maintainers: Scott Cunningham (scunning1975 on GitHub; Baylor University)
Compatibility:claude-code
The Claude Code toolkit of the author of Causal Inference: The Mixtape, and the most-cited economist skill repository in Velikov's AI-in-economics wiki (464 stars). Ten skills cover the empirical-economics workflow around a project rather than the estimation itself: an adversarial second referee for decks and for cross-language code replication, a many-agent bibliography audit that checks each citation's DOI and fields, a "blindspot" audit for problems and missed opportunities in empirical output, Beamer deck design and compilation built on the author's Rhetoric of Decks, figure-collision checks, careful chunked reading of PDFs, project and book scaffolding, and a warrant-first research task system. The repository has no licence file; its README grants informal permission to use the material freely. Because that permission does not spell out redistribution terms, RISE links to each skill rather than reproducing it.
Audits a .bib file by giving each citation to its own narrowly scoped agent, which confirms the DOI or URL and checks that every field belongs to the same paper, catching mismatched or invented references.
A quick collision check for figures, whether TikZ inside LaTeX or rendered images from R or Python, looking for overlapping labels and similar layout faults.
A research task-management system for causal-inference work that organises hypotheses, insights and decisions in separate folders and asks the researcher to state the warrant for each conjecture.
Scaffolds a book project: folder layout, a memoir-class LaTeX skeleton with a custom style, bibliography stub, CLAUDE.md, README and one file per chapter.
Downloads an academic PDF, splits it into short chunks and reads them in small batches, so that long papers are read closely rather than skimmed before a review or summary.
Reviews empirical output for what its author is too close to see, listing both problems hidden in plain sight and overlooked opportunities in the results.
An adversarial second-referee audit in two modes: reviewing a slide deck for rhetoric, visual quality and clean compilation, or re-implementing a paper's code in another language to check that the results replicate.
Builds a Beamer presentation end to end, designing a theme for the intended audience and restructuring existing material around the author's Rhetoric of Decks.