Warrant-first research GTD system. Manages the capture-clarify-organize-reflect-engage cycle for causal inference research. Scaffolds hypotheses/, insights/, decisions/ directories. Interrogates conjectures, files results, tracks binding decisions, checks pipeline freshness, drives the courtroom checklist.
Quick visual-collision check for figures — TikZ inside .tex files OR rendered .png/.jpg/.pdf figures from R/Python. Three checks only. (1) Bezier-curve label collisions via gap math. (2) Label-to-object whitespace. (3) Labels touching or running off the figure edge. These are the visual errors that compile cleanly — pdflatex never warns about them, ggsave never refuses to write them, so the agent that produced them does not know they are wrong. Use after generating any figure where visual correctness matters and you cannot eyeball it yourself.
Scaffold a new book project — folders, memoir-based LaTeX skeleton, custom style with voiced sidebars, bibliography stub, CLAUDE.md, README, and a chapter-per-file structure that converts cleanly to HTML later. Use at the start of a book project to get a compileable skeleton in one step. Parallel to /newproject and /bibcheck.
Many-agent bibliography audit. Verify each citation in a .bib file by spawning narrow-focus agents that confirm DOI/URL and cross-check that all fields belong to the same paper. Catches mixed-up entries (one paper's title with another's authors), wrong years, journal misattributions, and unverifiable references. Use when reviewing a manuscript's bibliography for accuracy before submission, after literature review, or when inheriting a .bib from a coauthor.
End-to-end beautiful Beamer deck creation. Designs an original Beamer theme tailored to a specific audience, restructures existing content via the Rhetoric of Decks (ethos / pathos / logos), generates figures and tables from R/Python/Stata code first, embeds code blocks in the deck, produces standalone walkthrough scripts, compiles to zero warnings, runs /tikz for visual collision cleanup, and dispatches a graphics-only audit agent for label and coordinate checks. Use when creating a presentation from scratch or restructuring existing content into a new beautiful deck.
Download, split, and deeply read academic PDFs. Use when asked to read, review, or summarize an academic paper. Splits PDFs into 4-page chunks, reads them in small batches, and produces structured reading notes — avoiding context window crashes and shallow comprehension.
Peripheral vision audit for empirical output. Finds what the author cannot see — problems hiding in plain sight (vices) and opportunities being overlooked (virtues). Inspired by Viktor Shklovsky's defamiliarization and conversations with Jason Fletcher. Use when output exists and interpretation is about to happen.
Systematic audit and review by Referee 2. Two modes — "deck" reviews slide presentations for rhetoric, visual quality, and compile cleanliness; "code" performs cross-language replication and econometric audit of empirical pipelines. Use when reviewing slides, auditing code, or verifying replication.