paper-poster

name	paper-poster
description	Turn a compiled paper into a conference poster: extract the poster-worthy content into a content plan, then hand it to /haipipe-display-poster to lay out (A0/A1 PDF + PPTX + SVG). Use when user says "做海报", "制作海报", "conference poster", "make poster", "生成poster", "poster session", or wants a poster for a paper. Owns the SELECTION (what a poster shows of a paper); the rendering belongs to display/.
argument-hint	[paper-directory-or-venue]
allowed-tools	Bash(*), Read, Write, Edit, Grep, Glob, Skill
metadata	{"version":"0.2.0","last_updated":"2026-07-24","summary":"Compiled paper → poster content plan → dispatch to haipipe-display-poster. Owns extraction; owns no layout."}

Paper → Poster

Build a poster for: $ARGUMENTS

What this skill owns

The selection, not the layout. A poster shows roughly a third of a paper; deciding which third — and turning dense prose into bullets and headline numbers — is a paper job, and it lives here. Once that plan exists, laying it out as a poster is a generic display job.

paper/main.tex + sections/*.tex + figures/
        │
        │  ← this skill: read, select, condense
        ▼
poster/poster-content-plan.md  +  poster/figures/     ← the handoff
        │
        │  ← /haipipe-display-poster: tcbposter, compile, review, export
        ▼
poster/poster.pdf · poster.pptx · poster.svg

The plan's shape is fixed by display/ref/content-plan-spec.md. Write it to that shape and the renderer needs nothing else — it never opens the paper.

Constants

• PAPER_DIR = paper/ — Directory containing the compiled paper (main.tex + sections/ + figures/).
• OUTPUT_DIR = poster/ — Where the plan, the figures, and the rendered poster land.
• VENUE = NeurIPS — Target venue (sets the renderer's colour scheme).
• POSTER_SIZE = A0 · ORIENTATION = landscape · COLUMNS = 4 Portrait A0 should use 3 columns, never 4 — text becomes too narrow to read.

These go into the plan's header, where they override the renderer's defaults.

Phase 0: Verify the paper

ls $PAPER_DIR/main.tex || ls $PAPER_DIR/main.pdf
ls $PAPER_DIR/sections/*.tex
ls $PAPER_DIR/figures/
mkdir -p $OUTPUT_DIR/figures

No compiled paper → stop and say so. This skill has nothing to extract from.

Phase 1: Extract the content plan

Read each section from $PAPER_DIR/sections/*.tex and select poster-appropriate content.

Extraction rules — a poster shows ~30-40% of the paper:

Paper Section	Poster Extraction	Target Length
Abstract	Skip — replace with 2-4 big-number stat callouts spanning all columns	0 words (numbers only)
Introduction	Motivation: 2-3 bullets + numbered contribution list (4 items)	120-160 words
Method	1 hero architecture figure + key equations + 3-5 bullets	80-120 words
Experiments	Dataset details + main result figures + numeric stat tables + ablation	150-200 words
Conclusion	3-4 key findings + 2-3 next steps	60-80 words
Related Work	Skip entirely — no space on a poster	0

Total target: 300-500 words of prose, excluding captions and callout numbers.

⚠️ No abstract paragraph on a poster. Replace it with a stat banner — 3-4 large headline numbers. This is the single highest-impact change for 60-second comprehension.

Figure selection — rank what's in $PAPER_DIR/figures/ and pick 3-5:

• Tier 1 (must): architecture / method overview, main results plot
• Tier 2 (if space): ablation chart, qualitative examples, experimental paradigm
• Tier 3 (skip): appendix figures, supplementary plots, tables-as-figures

Copy the chosen ones into $OUTPUT_DIR/figures/ (cp, never symlink).

Content-authoring rules:

• Never invent citations. Only references that exist in the paper's bibliography.
• Every number must be real — each callout and table cell traces to a result the paper reports.
• De-AI polish: no delve, pivotal, underscore, noteworthy, leverage, facilitate, harness.

Output: $OUTPUT_DIR/poster-content-plan.md, in the shape defined by display/ref/content-plan-spec.md — header (title, authors, venue, size, orientation, columns, link), ## Stat callouts, then one ### box per poster box carrying col:, optional figure: + caption:, optional stats:, and its bullets.

🚦 Checkpoint:

📋 Poster content plan ready:
- Title: [paper title]
- Venue: [VENUE] ([POSTER_SIZE] [ORIENTATION], [COLUMNS] columns)
- Stat callouts: [N]
- Boxes: Col1=[N], Col2=[N], Col3=[N], Col4=[N]
- Figures selected: [N] → copied into poster/figures/
- Prose word count: [N] (target 300-500)

Proceed to render? Or adjust the selection?

⛔ STOP HERE and wait for user response. This is the one decision that matters — once the plan is right, rendering is mechanical.

Phase 2: Hand off to the renderer

Skill(haipipe-display-poster)  with  PLAN=poster/poster-content-plan.md
                                     FIGURES_DIR=poster/figures/

From here the renderer owns everything: tcbposter layout, compile, visual review, and the PDF / PPTX / SVG exports. If it reports a box overrunning its column, fix the plan and re-render — do not fight the layout from this side.

Boundary

• ✅ This skill: what a poster shows of this paper, condensed into a content plan.
• ❌ Not this skill: how a poster looks — columns, cards, fonts, compile, export. That is haipipe-display-poster, and it serves any source, not just papers.