NeurIPS Workflow
Use this skill as the router for the NeurIPS depth pack. It helps choose the next skill and stage
the work so a manuscript does not arrive at submission with hidden policy or evidence gaps.
NeurIPS is a conference, not a journal: it has no standing editor-in-chief and no article-processing
charge. The rotating leadership is the per-edition General Chairs and Program Chairs (2026 General
Chairs: Hsuan-Tien Lin and Razvan Pascanu, verified 2026-06-22), and the cost model is registration
fees, not APCs — proceedings are open-access on OpenReview. Conference organizers rotate yearly, so
re-check the current CFP and committees page rather than carrying a name forward.
Stage map
- Topic and track fit: use
neurips-topic-selection.
- Paper shaping: use
neurips-writing-style, neurips-related-work, and neurips-experiments.
- Artifact and supplement planning: use
neurips-artifact-evaluation, neurips-reproducibility,
and neurips-supplementary.
- Final pre-submission audit: use
neurips-submission.
- Review diagnosis: use
neurips-review-process.
- Rebuttal and discussion: use
neurips-author-response.
- Acceptance and public release: use
neurips-camera-ready.
Operating principles
- Start official-source checks early; NeurIPS annual policies change.
- Keep track choice explicit. A main-track method paper, E&D paper, position paper, and MLRC paper
are not interchangeable.
- Write the checklist in parallel with the paper rather than at the end.
- Maintain one evidence table mapping every central claim to proof, experiment, data, artifact, and
limitation.
- Keep an anonymous review artifact and a de-anonymized public release artifact as separate states.
Deadline ladder to schedule backward from
NeurIPS spreads its gates across the year, so plan backward from the abstract deadline, not the
full-paper deadline. The exact dates move every cycle; reopen the current CFP and Main Track Handbook
for the live calendar.
| Gate |
What locks here |
Skill to finish before it |
| OpenReview profiles |
every author needs an active, complete profile |
neurips-submission |
| Abstract deadline |
title, abstract, and author list freeze (order may change after) |
neurips-topic-selection, neurips-related-work |
| Full-paper deadline |
PDF, checklist, and code/data ZIP submitted together |
neurips-experiments, neurips-supplementary |
| Reviews released |
nothing changes; you read and triage |
neurips-review-process |
| Author-response window |
rebuttal through OpenReview, no new paper version |
neurips-author-response |
| Decision |
accept (spotlight/poster) or reject |
neurips-camera-ready if accepted |
| Camera-ready |
de-anonymize, finalize checklist, public artifacts |
neurips-camera-ready, neurips-artifact-evaluation |
Track-fork early warning
Decide the track before drafting, because the Datasets & Benchmarks track, Position Papers track, and
MLRC/TMLR reproducibility route each have a distinct call, checklist emphasis, and review rubric. A
dataset paper forced into the main track, or a reproduction study that belongs in MLRC, wastes a full
cycle. If neurips-topic-selection flags a fork, resolve it before any other stage.
Worked vignette: where this router intervenes
A team has strong benchmark wins for a new retrieval method but no mechanism analysis and a thin
limitations paragraph. The router routes them to neurips-experiments (add ablations and error
analysis) and neurips-writing-style (calibrate the claim and the limitations) before
neurips-submission, rather than letting a leaderboard-only paper reach OpenReview where reviewers
predictably ask for the missing mechanism.
Output format
[Current stage] topic / drafting / pre-submission / review / rebuttal / camera-ready / reroute
[Next skill] <one NeurIPS skill>
[Official source to open] <CFP/handbook/template/OpenReview/checklist>
[Blocking gap] <policy/evidence/artifact/writing>
[Next action] <concrete task>
Source: brycewang-stanford/Awesome-Journal-Skills → NeurIPS-Skills/skills/neurips-workflow/SKILL.md
1---2name: neurips-workflow3description: Use when planning and sequencing a full NeurIPS manuscript workflow from topic and track selection through writing, the mandatory Paper Checklist, OpenReview submission, double-blind review, author response, decision, camera-ready, artifact release, and rerouting to the Datasets & Benchmarks track, a workshop, or MLRC/TMLR.4---567# NeurIPS Workflow89Use this skill as the router for the NeurIPS depth pack. It helps choose the next skill and stage10the work so a manuscript does not arrive at submission with hidden policy or evidence gaps.1112NeurIPS is a conference, not a journal: it has no standing editor-in-chief and no article-processing13charge. The rotating leadership is the per-edition General Chairs and Program Chairs (2026 General14Chairs: Hsuan-Tien Lin and Razvan Pascanu, verified 2026-06-22), and the cost model is registration15fees, not APCs — proceedings are open-access on OpenReview. Conference organizers rotate yearly, so16re-check the current CFP and committees page rather than carrying a name forward.1718## Stage map19201. Topic and track fit: use `neurips-topic-selection`.212. Paper shaping: use `neurips-writing-style`, `neurips-related-work`, and `neurips-experiments`.223. Artifact and supplement planning: use `neurips-artifact-evaluation`, `neurips-reproducibility`,23 and `neurips-supplementary`.244. Final pre-submission audit: use `neurips-submission`.255. Review diagnosis: use `neurips-review-process`.266. Rebuttal and discussion: use `neurips-author-response`.277. Acceptance and public release: use `neurips-camera-ready`.2829## Operating principles3031- Start official-source checks early; NeurIPS annual policies change.32- Keep track choice explicit. A main-track method paper, E&D paper, position paper, and MLRC paper33 are not interchangeable.34- Write the checklist in parallel with the paper rather than at the end.35- Maintain one evidence table mapping every central claim to proof, experiment, data, artifact, and36 limitation.37- Keep an anonymous review artifact and a de-anonymized public release artifact as separate states.3839## Deadline ladder to schedule backward from4041NeurIPS spreads its gates across the year, so plan backward from the abstract deadline, not the42full-paper deadline. The exact dates move every cycle; reopen the current CFP and Main Track Handbook43for the live calendar.4445| Gate | What locks here | Skill to finish before it |46| --- | --- | --- |47| OpenReview profiles | every author needs an active, complete profile | `neurips-submission` |48| Abstract deadline | title, abstract, and author list freeze (order may change after) | `neurips-topic-selection`, `neurips-related-work` |49| Full-paper deadline | PDF, checklist, and code/data ZIP submitted together | `neurips-experiments`, `neurips-supplementary` |50| Reviews released | nothing changes; you read and triage | `neurips-review-process` |51| Author-response window | rebuttal through OpenReview, no new paper version | `neurips-author-response` |52| Decision | accept (spotlight/poster) or reject | `neurips-camera-ready` if accepted |53| Camera-ready | de-anonymize, finalize checklist, public artifacts | `neurips-camera-ready`, `neurips-artifact-evaluation` |5455## Track-fork early warning5657Decide the track before drafting, because the Datasets & Benchmarks track, Position Papers track, and58MLRC/TMLR reproducibility route each have a distinct call, checklist emphasis, and review rubric. A59dataset paper forced into the main track, or a reproduction study that belongs in MLRC, wastes a full60cycle. If `neurips-topic-selection` flags a fork, resolve it before any other stage.6162## Worked vignette: where this router intervenes6364A team has strong benchmark wins for a new retrieval method but no mechanism analysis and a thin65limitations paragraph. The router routes them to `neurips-experiments` (add ablations and error66analysis) and `neurips-writing-style` (calibrate the claim and the limitations) before67`neurips-submission`, rather than letting a leaderboard-only paper reach OpenReview where reviewers68predictably ask for the missing mechanism.6970## Output format7172```text73[Current stage] topic / drafting / pre-submission / review / rebuttal / camera-ready / reroute74[Next skill] <one NeurIPS skill>75[Official source to open] <CFP/handbook/template/OpenReview/checklist>76[Blocking gap] <policy/evidence/artifact/writing>77[Next action] <concrete task>78```7980---8182**Source:** [`brycewang-stanford/Awesome-Journal-Skills`](https://github.com/brycewang-stanford/Awesome-Journal-Skills) → `NeurIPS-Skills/skills/neurips-workflow/SKILL.md`
thedixitjain