Research Presentation

---
name: research-presentation
description: The standard for how research is presented — structuring a research presentation, slide deck, research update, mentor or team meeting, a report or an artifact page. Use when building, structuring or critiquing slides, a research update, a findings page, or a figure destined for a deck.
---

# Research Presentation

The standard lives in **`~/.claude/checklists/presentation.md`**. Read it now; it is the content, and this file only routes to it.

It covers the unit of claim + figure + elaboration and the composition test of reading only headings and first sentences; numbers escalating to a plot and stopping there; headings that assert; terminology, undefined jargon and the words that sound standard and are not; bolded bullet takeaways for skimming reviewers, and where that pulls against `writing.md`; attention management, progressive disclosure and collapsed rows that carry their outcome; the slide sequence — summary slide, agenda, experiments before results, most important results first, one message per slide, backup slides, concrete discussion points, one consistent project story; and what makes a figure readable in seconds. The file doubles as a review pass: judge each item Good / Needs improvement / Missing, and for every problem say why it matters and propose a concrete improvement.

Critique and improve proactively whenever slides, a research update or a plot come past — do not wait to be asked.

## Local references the checklist does not carry

- **`references/paper-figures.md`** — publication figures for ICML/NeurIPS/ICLR: single- and double-column sizes in inches, font sizes that survive 50% scaling, `text.usetex` integration, PDF export and the pre-submission checklist (vector, embedded fonts, grayscale-distinguishable). Load this only for a paper figure, never for slides. It carries no colours of its own — every hex comes from `house-plots` and `lib/plotting/palette.py`.
- **`references/templates.md`** — the research-update email skeleton (outcome → tried → result against baseline → decision needed → next steps) and the matching slide skeleton.

`GUIDE.md`, `references/common-mistakes.md` and `references/visualization-guide.md` predate the checklist and now restate it; prefer `~/.claude/checklists/presentation.md` over all three.

## An exported deck can eat the whole context window

Never read an exported PDF or a long deck in main context. Delegate it: a subagent reads the PDF and reports the issues, and the `slidev` skill delegates its own PDF analysis the same way. Creating slides directly in main context is fine; reviewing the export is not.

## Reach for a neighbour instead when

- the question is **Slidev tooling** — building the deck, export, overflow, or a single slide that is cut off or blank: the `slidev` skill
- the question is **whether the evidence holds** rather than how it looks — intervals, nulls, chance correction, slicing, provenance: `~/.claude/checklists/results-analysis.md`
- the question is about **sentences and paragraphs** rather than the page or deck: `~/.claude/checklists/writing.md`
- you are **drawing the chart**: `house-plots` for matplotlib in papers and decks, the built-in `dataviz` for native SVG on an artifact page
- you are building the **page** — transcripts, collapsible units, the annotation layer, `md2artifact`: the `artifact-writing` skill
- the draft is finished and the risk is being **misread**: the `reduce-ambiguity` skill

## Why this file is thin

Five checklists hold the content; skills route to them. Restating a standard in every skill that touches it is how five skills came to say overlapping things about clarity, with edits landing in one copy and not the others. See `~/.claude/checklists/README.md`.