Academic PPTX

Academic presentation builder

Create a .pptx that looks authored by a careful academic presenter: evidence first, generous whitespace, restrained color, and no decorative filler.

Compose with the Presentations skill

Use the installed presentations:Presentations skill for PowerPoint creation, dependency loading, rendering, slide inspection, and file delivery. Read it before building. Its artifact and QA requirements still apply; this skill supplies the academic content and visual system.

The Presentations skill's hard requirements, implementation route, template precedence, and verification workflow always win. This skill narrows the academic content, evidence, and visual-role decisions inside that workflow; it does not create a second slide-authoring route.

Establish the brief

Infer what is safe from the request and source material. Ask only for missing information that would materially change the deck:

• topic and purpose;
• audience and expected background;
• duration or target slide count;
• presenter, affiliation, date, and venue if wanted;
• source papers or guideline material;
• institutional branding, if required.

Do not invent study results, citations, presenter details, or institutional branding. Use current, authoritative sources when the requested content is time-sensitive. Keep a source ledger while drafting so every numerical or clinical claim can be traced.

Build the argument before the slides

Write a compact slide outline with message-based titles. A reader should understand the argument by reading only the titles. Generic labels such as Background, Methods, Results, and Discussion are not acceptable slide titles, except for an agenda or closing slide.

Typical sequence:

1. title;
2. why the question matters;
3. what was known and what remained uncertain;
4. methods or evidence frame;
5. main results, usually one conclusion per slide;
6. interpretation and limitations;
7. implications or practice/research consequences;
8. three to five take-home messages;
9. acknowledgments or questions, if useful.

For a 10–15 minute talk, omit a separate agenda unless it genuinely helps orientation. For a journal club, include study design, population, intervention/exposure, comparator, outcomes, effect estimates, bias/limitations, applicability, and a bottom-line appraisal.

Use the Presentations implementation route

Create the deck only with the current presentations:Presentations artifact-tool workflow. Do not use a legacy artifact API or a second slide library. Follow the Presentations skill's route selection exactly: a user-supplied reference wins; explicit user styling uses the custom-formatting route; otherwise use its bundled Codex Grid composition reference.

Use the Presentations skill's wide-slide setup and verify that the exported page is a true 16:9 layout equivalent to 13.333 × 7.5 inches. Do not substitute a mislabeled custom page size.

Use these wide-slide bounds as defaults:

• edge margin: at least 0.65 inches;
• title box: x 0.67, y 0.42, w 12.0, h 0.9;
• content begins at y >= 1.45 below the one-line title box;
• citation/footer zone: x 0.67, y 7.08, w 12.0, h 0.22;
• reserve the bottom 0.35 inches for citations or source notes;
• never place content outside x 0.5–12.83 or y 0.3–7.2.

Every slide title must remain on one line. Shorten the title or change the composition when it wraps; do not shrink it below the Presentations minimum or expand it to a second line. Reflow the slide when content does not fit.

Academic visual system

When no user reference, brand system, or selected Presentations route supplies a controlling palette, use white backgrounds and this role-based palette:

const C = {
  bg: "FFFFFF",
  text: "1F2937",
  muted: "667085",
  faint: "D0D5DD",
  panel: "F5F7FA",
  primary: "1D4E74",
  secondary: "2E7D8A",
  primaryFaint: "E3ECF3",
  categorical: ["1D4E74", "2E7D8A", "B08400", "7A5195"],
  good: "2F7D4F",
  caution: "B45309",
  harm: "9B2C2C"
};

Every non-neutral color must have a job:

• primary: deck identity and the most important series or number;
• secondary: the genuine counterpart in a two-part comparison;
• categorical: chart series only, with at most four hues;
• semantic colors: true favorable/caution/harm meaning only, with a legend or note.

Use charcoal for prose. Do not add colors merely for variety. Institutional colors may replace the primary/secondary pair, but preserve accessible contrast and semantic meaning.

When no user template supplies typography, use these defaults; they meet the Presentations skill's hard minimums:

Use a widely available typeface such as Arial, Aptos, Calibri, or Helvetica. Keep body text left aligned. Center only a deliberate title or large-stat composition.

Match visual form to evidence

• Use one chart per results slide when a chart is the evidence.
• Use artifact-tool native editable charts for quantitative evidence where practical.
• Start axes at a defensible value and disclose truncation; never exaggerate differences.
• Direct-label important series when legible; otherwise use a restrained legend.
• Put a one-sentence interpretation beneath a chart. It should say what the pattern means, not repeat the title.
• Use tables for exact multi-study comparisons. Highlight one row or column at most, and keep borders light.
• Follow the Presentations visual-asset contract: do not draw images in Python or create programmatic vector illustrations. Use native shapes only for a simple diagram that materially improves understanding, and use the Presentations-prescribed visual route for complex or illustrative diagrams.
• Use source figures only when permitted and necessary. Preserve aspect ratio, crop intentionally, and cite the source on the same slide.
• Add alt text or meaningful descriptions when the authoring tool supports them.

Do not use decorative medical icons, stock clinical photos, gradients, 3D effects, glow, full-width color bars, colored rules under every title, logos on every slide, or a border around every text box.

Cite claims on the slide

Add a compact source line to each slide that contains external evidence. Use enough information to identify the source consistently—for example, first author, journal, year, and DOI/PMID or a short URL. Put complete references on a final references slide only when the audience or venue expects it.

Never cite a source that does not support the exact claim. Distinguish reported findings from your interpretation. Flag preprints, subgroup analyses, indirect evidence, and low-certainty evidence clearly.

Write useful speaker notes

Add notes to every substantive slide. Write cues for the presenter, not a polished narration:

• vary note length;
• use fragments and reminders where natural;
• include pronunciation, timing, likely questions, or caveats when useful;
• avoid opening every note with This slide shows;
• avoid a scripted transition at the end of every slide;
• never place unsupported claims in notes.

Example: Pause on the absolute difference. The relative effect sounds larger. If asked about crossover, point to the supplement.

Required build and QA loop

1. Use the Presentations skill to create the editable .pptx with a true wide layout.
2. Render the deck using the Presentations skill's current renderer; resolve tools from the installed skill and workspace rather than assuming a fixed path or local binary.
3. Inspect every rendered slide, preferably as both a montage and individual full-size pages.
4. Check for clipped text, overflow, overlaps, title collisions, unreadable labels, distorted figures, low contrast, and inconsistent margins.
5. Check evidence fidelity: numbers, units, denominators, CI bounds, legends, citations, and direction of effect.
6. Fix every material defect, render again, and inspect the changed slides. Run any overflow or overlap diagnostic supplied by the Presentations skill.
7. Save the final .pptx in the task's user-facing output location and return a clickable file link. Include a PDF only if the user requested one or it improves review.

Final acceptance checks

• Page size is 16:9 wide, not a mislabeled custom size.
• Titles state conclusions rather than section names.
• No slide contains invented evidence or an unverifiable citation.
• Body text is normally at least 18 pt; exceptions are limited to tables, chart labels, and citations.
• No colored element lacks a defined role.
• Layouts vary with the content without becoming stylistically inconsistent.
• All substantive slides have useful notes.
• Every rendered slide has been visually inspected after the final material change.