skills/13-scunning1975-MixtapeTools/skills/compiledeck/SKILL.md
Create and compile beautiful Beamer presentations following the Rhetoric of Decks philosophy. Use when making slides, creating decks, or compiling .tex presentation files.
npx skillsauth add brycewang-stanford/Awesome-Agent-Skills-for-Empirical-Research compiledeckInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Create beautiful Beamer presentations. This skill enforces the Rhetoric of Decks philosophy through concrete, measurable rules — not aspirational guidance.
Before touching anything, answer three questions:
Q1: New deck or editing existing?
.tex file first. Understand its palette, structure, and style before changing anything.Q2: Who is the audience?
See domain_patterns.md in this skill directory for detailed guidance per audience type.
Q3: What's the tone?
palette_reference.md for inspiration.These are not guidelines. They are constraints.
If a slide has two ideas, split it. No exceptions.
If someone reads only the slide titles in sequence, they should understand the entire argument.
Bullets are a confession that you haven't found the structure. Look for:
Exception: genuinely parallel items (e.g., a list of axioms in a teaching deck).
Empty space on a slide signals that you chose carefully. A full slide signals anxiety.
Not "try to fix." Zero. See the Compile Loop below.
For new decks, use this template. It is extracted from real decks that work.
\documentclass[aspectratio=169,11pt]{beamer}
% Packages
\usepackage[utf8]{inputenc}
\usepackage[T1]{fontenc}
\usepackage{lmodern}
\usepackage{microtype}
\usepackage{amsmath,amssymb}
\usepackage{booktabs}
\usepackage{graphicx}
\usepackage{tikz}
\usepackage{pgfplots}
\usepackage{xcolor}
\usepackage{ragged2e}
\usepackage{colortbl}
\usepackage{array}
\usepackage{hyperref}
\usepackage{adjustbox}
\usetikzlibrary{shapes.geometric, arrows.meta, positioning, calc,
backgrounds, decorations.pathreplacing, shadows, fadings}
\pgfplotsset{compat=1.18}
% --- Color Palette: Warm Professional ---
\definecolor{DeepNavy}{HTML}{2E4057}
\definecolor{Teal}{HTML}{048A81}
\definecolor{WarmOrange}{HTML}{E85D04}
\definecolor{SoftPurple}{HTML}{9D4EDD}
\definecolor{WarmGray}{HTML}{6C757D}
\definecolor{LightGray}{HTML}{E9ECEF}
\definecolor{Cream}{HTML}{FBF8F1}
\definecolor{DeepRed}{HTML}{D62828}
\definecolor{Gold}{HTML}{D4A03A}
\definecolor{SoftWhite}{HTML}{FAFAFA}
% Beamer colors
\setbeamercolor{normal text}{fg=DeepNavy, bg=SoftWhite}
\setbeamercolor{structure}{fg=DeepNavy}
\setbeamercolor{alerted text}{fg=DeepRed}
\setbeamercolor{example text}{fg=Teal}
\setbeamercolor{frametitle}{fg=DeepNavy, bg=SoftWhite}
\setbeamercolor{title}{fg=DeepNavy}
\setbeamercolor{subtitle}{fg=WarmGray}
\setbeamercolor{block title}{bg=DeepNavy, fg=SoftWhite}
\setbeamercolor{block body}{bg=LightGray, fg=DeepNavy}
\setbeamercolor{itemize item}{fg=WarmOrange}
\setbeamercolor{itemize subitem}{fg=Gold}
% Fonts
\usefonttheme{professionalfonts}
\setbeamerfont{title}{size=\huge, series=\bfseries}
\setbeamerfont{frametitle}{size=\Large, series=\bfseries}
% Strip chrome
\setbeamertemplate{navigation symbols}{}
\setbeamertemplate{headline}{}
% Minimal footline
\setbeamertemplate{footline}{%
\hfill
\begin{beamercolorbox}[wd=3cm,ht=2.5ex,dp=1ex,right,rightskip=0.6cm]{page number}
\usebeamercolor[fg]{WarmGray}\scriptsize\insertframenumber
\end{beamercolorbox}
\vspace{0.3cm}
}
% Bullets
\setbeamertemplate{itemize item}{\footnotesize\raisebox{0.3ex}{\tikz\fill[WarmOrange] (0,0) circle (0.45ex);}}
\setbeamertemplate{itemize subitem}{\footnotesize\raisebox{0.3ex}{\tikz\fill[Gold] (0,0) circle (0.35ex);}}
% Custom commands
\newcommand{\transitionslide}[2]{%
{\setbeamercolor{normal text}{fg=SoftWhite,bg=DeepNavy}
\begin{frame}[plain]\vfill\begin{center}
{\huge\bfseries\textcolor{Gold}{#1}}\\[0.6cm]
{\large\textcolor{LightGray}{#2}}
\end{center}\vfill\end{frame}}
}
\graphicspath{{figures/}}
For alternative palettes, see palette_reference.md in this skill directory.
NOT "Motivation" with bullet points. Instead:
Use \transitionslide{Section Title}{Subtitle} for dark-background section dividers.
NOT "Questions?" or "Thank you." Instead:
Present the strongest objection to your argument, then respond to it. This builds credibility. Format: "A skeptic would say..." followed by your response.
Read tikz_rules.md in this skill directory BEFORE creating any TikZ figure. The rules are measurement-based — they prevent text from overlapping arrows, boxes, and other objects.
Critical rule preview: Before placing a label between two nodes, CALCULATE the edge-to-edge gap and compare it to the estimated text width. If the text is wider than the gap minus 0.6cm padding, the label WILL collide. Move it above/below or shorten the text.
After EVERY edit to a .tex file:
pdflatex -interaction=nonstopmode [file].tex
grep "^!" [file].log
If any: fix, recompile, repeat.
grep -cE "Overfull|Underfull" [file].log
This must return 0. For EACH warning:
| Warning | Fix |
|---------|-----|
| Overfull \hbox | Rephrase shorter, use \adjustbox, add @{} to table columns |
| Underfull \hbox | Rephrase or adjust paragraph breaks |
| Overfull \vbox | Split the slide or reduce \vspace, tighten content |
| Underfull \vbox | Add \vfill or adjust spacing |
Even 0.5pt overfull must be fixed. Do NOT proceed until the count is 0.
grep -i "font" [file].log | grep -i "warning"
TikZ errors don't trigger compile warnings. You must catch them yourself. Follow this order — curves first, everything else second:
Pass 1 — Find and fix all Bézier curves:
grep -n "bend" [file].tex
For EACH curved arrow found:
(chord / 2) × tan(bend_angle / 2)Pass 2 — Everything else:
above, below, left, or right?Pass 3 — Open the PDF and visually confirm.
Full rules and formulas are in tikz_rules.md. This verification runs on ALL TikZ figures in the deck, not just the one you just edited.
open [file].pdf
Every deck tells a story. The structure depends on context, but the principle is universal:
Act I — Setup: Establish the problem or question. Make the audience feel it. Act II — Development: Present evidence, analysis, investigation. Build the logical case. Act III — Resolution: Deliver the insight. State implications. Give the audience something to do or remember.
Within this arc, apply the Pyramid Principle: lead with the conclusion, then support it. Academic audiences expect this. Do not build suspense — state the finding, then prove it.
For domain-specific structures, see domain_patterns.md.
When a deck needs data visualizations:
generate_figures.py) using matplotlibfigures/ subdirectoryfig.set_facecolor()ax.text() near patches (FancyBboxPatch, Circle, Rectangle), compute the patch boundary and verify text has ≥0.4cm clearance. See tikz_rules.md Pass 4.va='center' for both at symmetric offsets — multi-line text extends further from its anchor than single-line, creating visual asymmetry. Instead, anchor outward from center: upper element uses va='bottom' at center + gap, lower uses va='top' at center - gap. See tikz_rules.md Pass 4.arc3), compute the actual Bézier curve position using the arc3 control point formula (cx = mid_x + rad*dy, cy = mid_y - rad*dx), then offset labels perpendicular to the computed curve position. Never guess where curves pass. Use a white-background bbox on labels for readability. See tikz_rules.md Pass 4 for the full formula and workflow.| Principle | Rule | |-----------|------| | Ideas per slide | ONE | | Title style | Assertion, not label | | Minimum font | 24pt (18pt floor) | | Bullets | Avoid — find the structure | | White space | Confidence, not waste | | Charts | One message, direct labels | | Compile target | Zero warnings | | Opening slide | Puzzle, surprise, or provocation | | Closing slide | One memorable takeaway | | TikZ labels | Calculate the gap (see tikz_rules.md) | | Color palette | 10+ colors, warm by default | | Narrative | Setup, Development, Resolution |
All in this skill directory (~/.claude/skills/compiledeck/):
For the complete essay behind these principles:
~/MixtapeTools/presentations/rhetoric_of_decks.md
~/MixtapeTools/presentations/rhetoric_of_decks_full_essay.md
This skill distills those essays into operational rules. You don't need to re-read them — just follow the rules above.
tools
Recommend AND run open-source AI tools, agents, Claude Code / Codex skills, and MCP servers for any stage of a literature review — searching, reading, extracting, synthesizing, screening, citation-checking, and paper writing. Use when the user asks "what tool should I use to..." OR "install/run/use <tool> to ..." for research/lit-review work: automating a survey or related-work section, PDF→Markdown extraction for LLMs (MinerU/marker/docling), PRISMA / systematic review (ASReview), citation-backed Q&A over PDFs (PaperQA2), wiring papers into Claude/Cursor via MCP (arxiv/paper-search/zotero servers), or chatting with a Zotero library. Ships a launcher (scripts/litrun.py) that installs each tool in an isolated venv and runs it. Curated catalog of 70+ vetted projects. 支持中英文(用于「文献综述工具选型」与「一键安装/运行」)。
development
Route empirical-research requests through the Auto-Empirical Research Skills catalog when this whole repository is installed as one skill in Codex, CodeBuddy, Claude Code, or another IDE. Use to choose and load the right vendored AERS skill for causal inference, econometrics, replication, data acquisition, manuscript writing, peer review and referee responses, citation checking, de-AIGC editing, or full empirical-paper workflows without reading the entire repository at once.
documentation
Use when the project collects primary data or runs a field, lab, or survey experiment, before the intervention begins — write the pre-analysis plan, size the sample from a power calculation, and register with the AEA RCT Registry. Apply after the design is chosen in aer-identification and before any outcome data are seen.
tools
Guide economists to authoritative data sources with explicit, confirmed data specifications before retrieval; interfaces with Playwright MCP to navigate portals and extract real data, not articles about data.