brycewang-stanford/awesome-journal-skills/aistats-supplementary
- community
Checking that the SKILL.md file is still in the repository.
Use when preparing AISTATS supplementary material, appendices, proof details, code/data archives, simulation scripts, additional tables, and anonymized artifacts under deadline, size, anonymity, and reviewer-discretion constraints, including how to split a theory-plus-experiments paper between body and supplement.
Install
npx vskill@latest install brycewang-stanford/awesome-journal-skills/aistats-supplementarySKILL.md
AISTATS Supplementary
Use this when assembling AISTATS supplementary material. The supplement can support the paper, but the main submission must remain understandable without it.
Supplement structure
- Put theorem proofs, derivations, extra ablations, simulation details, dataset
documentation, and robustness tables in a clean appendix or supplementary manuscript.
- Put executable assets in a separate archive when current OpenReview forms allow it.
- Respect the current supplementary deadline. AISTATS 2026 placed supplementary-material
submission after the paper deadline but before review.
- Keep all supplementary files double-blind: no authors, institutions, acknowledgements,
private paths, repository owners, grants, or identifying metadata.
- Do not use the supplement to hide essential motivation, core method description, main
theorem statements, or primary empirical results.
- Verify archives open on a clean machine and contain no credentials, cache directories,
large irrelevant files, or hidden OS metadata.
Appendix architecture for proofs
- Order appendix sections to mirror main-text theorem order; statistician reviewers navigate
by theorem number, not by page.
- Restate each theorem before its proof so the appendix reads standalone without flipping
back to the two-column body.
- Standard partition: notation and assumptions, main proofs, auxiliary lemmas, additional
simulations, then real-data details.
- Keep one proof-sketch paragraph per major theorem in the body itself; an AISTATS submission
whose entire argument lives in the supplement reads as unreviewable within the page limit.
- Cross-reference every appendix table and lemma from the body at least once; orphaned
supplement material is invisible to reviewers and wasted under reviewer discretion.
- If the cycle allows appendices inside the main PDF after the references, confirm against
the current instructions whether that or a separate file upload is expected.
What gets opened first
| Supplement item | Inspection likelihood | Practical implication | |
|---|---|---|---|
| Proof of the headline theorem | High | Polish it to main-text standard | |
| Extra ablation and robustness tables | Medium | Reference each one from the body | |
| Code archive | Variable, reviewer discretion | The README must orient a reader in one minute | |
| Raw experiment logs | Low | Include for completeness, never rely on them |
Vignette: splitting an optimization paper
A submission proving a convergence guarantee for a stochastic mirror-descent variant, plus benchmark plots: the body keeps the theorem, the key lemma, the proof roadmap, and two decision-critical figures; the appendix holds full proofs and step-size sensitivity grids; the archive carries the seeded experiment runner. Nothing decision-critical lives only in the archive, because archive inspection at AISTATS is discretionary.
Output format
Security
- Security scanPassed, 90 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingAISTATS-Skills/skills/aistats-supplementary/SKILL.md
Popularity
- 7-day trend score
- 5
Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/brycewang-stanford/awesome-journal-skills/aistats-supplementary)