Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Post-acceptance conference preparation workflow covering presentation slides, academic posters, and promotion content. Triggers on "scholar publish", "conference preparation", "prepare presentation", "create poster", "write promotion", "post-acceptance".
.claude/skills/catlog22-scholar-publish/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | 90% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 118% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 102% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 33% | 0% |
| case-14 | ✗→✓ | ▲ Improved | 82% | 0% |
Post-acceptance conference preparation workflow that helps researchers create presentation materials, academic posters, and promotional content for accepted papers.
.workflow/codebase/ARCHITECTURE.md exists, read for project contextmaestro load --type spec --category coding — load coding conventionsmaestro search "academic writing research paper" --json — top 5 entries as prior context┌──────────────────────────────────────────────────────────────┐
│ /scholar-publish │
│ Orchestrator: Preference Collection + Selective Phase Dispatch │
└──────────────────────┬───────────────────────────────────────┘
│
┌─────────────┼─────────────┐
↓ ↓ ↓
┌───────────┐ ┌───────────┐ ┌───────────┐
│ Phase 1 │ │ Phase 2 │ │ Phase 3 │
│ Presenta- │ │ Poster │ │ Promotion │
│ tion │ │ Design │ │ Content │
└─────┬─────┘ └─────┬─────┘ └─────┬─────┘
↓ ↓ ↓
presentation- poster- promotion-
outline.md outline.md content.mdData Flow:
User Input: accepted paper + preferences (outputs, duration, poster size)
→ Phase 1 (presentation): paper → slide outline + timing plan + Q&A prep
→ Phase 2 (poster): paper → poster layout + design specs + QR code plan
→ Phase 3 (promotion): paper → Twitter thread + LinkedIn post + blog draft
Each phase runs independently based on user selection.Collect workflow preferences via AskUserQuestion before dispatching to phases:
Step 1: Identify paper context
Ask: "Please provide the accepted paper (file path, summary, or key details):
- Paper title
- Conference/venue name
- Key contributions (2-3 bullet points)
- Co-authors (for tagging in promotion)"
Step 2: Select outputs to generate
AskUserQuestion:
question: "Which outputs would you like to generate?"
options:
- "Presentation slides outline" → enablePresentation = true
- "Academic poster design" → enablePoster = true
- "Promotion content (Twitter/LinkedIn/blog)" → enablePromotion = true
- "All of the above" → enableAll = true
Step 3: Conditional preferences (based on selection)
IF enablePresentation:
Ask: "Talk duration?" options: ["15 minutes", "20 minutes", "30 minutes", "Other"]
IF enablePoster:
Ask: "Poster format?" options: ["Portrait (24x36in / A0)", "Landscape (36x24in / A0)", "Check with conference"]
IF enablePromotion:
Ask: "Which platforms?" options: ["Twitter/X thread", "LinkedIn post", "Blog post", "All platforms"]Preference Variables (passed to phases):
paperContext: title, venue, contributions, authorsenablePresentation, enablePoster, enablePromotiontalkDuration: 15 | 20 | 30 (minutes)posterFormat: portrait | landscapepromotionPlatforms: twitter | linkedin | blog | all> COMPACT DIRECTIVE: Context compression MUST check TodoWrite phase status. > The phase currently marked in_progress is the active execution phase - preserve its FULL content. > Only compress phases marked completed or pending.
Parse user input and collect preferences (described above). Convert free-text paper description to structured format:
PAPER_TITLE: [title]
VENUE: [conference/journal name]
CONTRIBUTIONS: [key points]
AUTHORS: [author list]Then dispatch to selected phases.
Ref: phases/01-presentation.md
Create a structured presentation outline with slide-by-slide content, timing plan, visual design guidance, and Q&A preparation.
Output: presentation-outline.md
Ref: phases/02-poster.md
Design a poster layout with section placement, typography specs, visual hierarchy, and print-ready guidelines.
Output: poster-outline.md
Ref: phases/03-promotion.md
Generate platform-specific promotion content: Twitter/X thread, LinkedIn post, and blog draft.
Output: promotion-content.md
Phase Reference Documents (read on-demand when phase executes):
| Phase | Document | Purpose | Compact | |-------|----------|---------|---------| | 1 | phases/01-presentation.md | Slide outline creation | TodoWrite driven | | 2 | phases/02-poster.md | Poster layout design | TodoWrite driven | | 3 | phases/03-promotion.md | Multi-platform promotion | TodoWrite driven |
Compact Rules:
in_progress -> preserve full content, do not compresscompleted -> can compress to summarypaperContext (from user input)
│
├─→ Phase 1: paperContext + talkDuration → presentation-outline.md
│ └─ slide count, timing, structure, visual design, Q&A prep
│
├─→ Phase 2: paperContext + posterFormat → poster-outline.md
│ └─ layout, typography, sections, QR code, print specs
│
└─→ Phase 3: paperContext + promotionPlatforms → promotion-content.md
└─ Twitter thread, LinkedIn post, blog draftPhase starts:
→ Sub-tasks ATTACHED to TodoWrite (in_progress + pending)
→ Execute sub-tasks sequentially within the phase
Phase ends:
→ Sub-tasks COLLAPSED to completed summary
→ Next selected phase begins (or workflow completes)Example TodoWrite lifecycle:
[in_progress] Generate presentation outline
[in_progress] Extract key messages from paper
[pending] Create slide structure
[pending] Add timing plan
[pending] Write Q&A preparation notes
→ After completion:
[completed] Generate presentation outline (12 slides, 20min talk)Before each phase:
After each phase:
After all phases:
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-05 | pass→fail | 18,837 | 20,656 | +10% | 1 | 1 | 0% | 2,940 | 5,178 | +76% | 0 | 0 | — |
case-06 | pass→pass | 14,136 | 15,771 | +12% | 1 | 1 | 0% | 2,131 | 4,300 | +102% | 0 | 0 | — |
case-07 | fail→pass | 7,814 | 1,703 | -78% | 1 | 1 | 0% | 1,124 | 2,135 | +90% | 0 | 0 | — |
case-01 | fail→fail | 9,577 | 6,768 | -29% | 1 | 1 | 0% | 1,495 | 3,032 | +103% | 0 | 0 | — |
case-02 | fail→fail | 10,707 | 12,428 | +16% | 1 | 1 | 0% | 1,631 | 2,798 | +72% | 0 | 0 | — |
case-03 | fail→fail | 10,984 | 6,113 | -44% | 1 | 1 | 0% | 1,698 | 2,909 | +71% | 0 | 0 | — |
case-04 | pass→pass | 18,178 | 22,323 | +23% | 1 | 1 | 0% | 2,694 | 5,193 | +93% | 0 | 0 | — |
case-08 | fail→pass | 6,401 | 1,356 | -79% | 1 | 1 | 0% | 975 | 2,122 | +118% | 0 | 0 | — |
case-09 | fail→pass | 7,656 | 2,354 | -69% | 1 | 1 | 0% | 1,142 | 2,305 | +102% | 0 | 0 | — |
case-10 | pass→pass | 15,453 | 12,442 | -19% | 1 | 1 | 0% | 2,376 | 3,700 | +56% | 0 | 0 | — |
case-11 | pass→pass | 9,752 | 2,113 | -78% | 1 | 1 | 0% | 1,483 | 2,273 | +53% | 0 | 0 | — |
case-12 | fail→pass | 11,920 | 2,862 | -76% | 1 | 1 | 0% | 1,805 | 2,397 | +33% | 0 | 0 | — |
case-13 | pass→pass | 15,823 | 5,296 | -67% | 1 | 1 | 0% | 2,355 | 2,731 | +16% | 0 | 0 | — |
case-14 | fail→pass | 7,908 | 1,984 | -75% | 1 | 1 | 0% | 1,234 | 2,240 | +82% | 0 | 0 | — |
case-15 | fail→pass | 11,652 | 3,663 | -69% | 1 | 1 | 0% | 1,636 | 2,466 | +51% | 0 | 0 | — |
case-16 | pass→pass | 16,753 | 3,840 | -77% | 1 | 1 | 0% | 2,353 | 2,542 | +8% | 0 | 0 | — |
case-17 | pass→pass | 13,539 | 7,759 | -43% | 1 | 1 | 0% | 1,758 | 2,967 | +69% | 0 | 0 | — |
case-18 | fail→pass | 10,296 | 4,103 | -60% | 1 | 1 | 0% | 1,395 | 2,514 | +80% | 0 | 0 | — |
case-19 | pass→pass | 12,483 | 7,746 | -38% | 1 | 1 | 0% | 1,819 | 3,123 | +72% | 0 | 0 | — |
case-20 | fail→pass | 13,676 | 5,943 | -57% | 1 | 1 | 0% | 1,851 | 2,913 | +57% | 0 | 0 | — |
case-21 | fail→pass | 4,711 | 1,594 | -66% | 1 | 1 | 0% | 692 | 2,162 | +212% | 0 | 0 | — |
case-22 | pass→pass | 9,792 | 4,406 | -55% | 1 | 1 | 0% | 1,402 | 2,576 | +84% | 0 | 0 | — |
case-23 | fail→pass | 12,117 | 8,464 | -30% | 1 | 1 | 0% | 1,748 | 3,242 | +85% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 23 cases were attempted. The headline lift of +39 percentage points is the difference between those two pass rates over the 23 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.