Use when drafting AISTATS author responses or author-reviewer discussion replies under OpenReview, covering text-only discussion, no-link guidance, no revised-paper upload, anonymity requirements, statistician-reviewer pushback patterns, and decision-focused clarification strategy for theory-plus-experiments papers.
SkillScore91Maintained
Automatic
◆Research pack
Last verified 8/15/2026
What it changes
Passed 20 of 22 test cases with the skill, and 10 of 22 without it.
Without skill45%
With skill91%
Checks it passed that the model alone missed
The draft response omits external URLs and web links.
The Decision dimension is set to clarity or correctness.
The Decision dimension is set to statistical validity or correctness.
and 11 more in the full report
11 fixed9 passed either way1 still failing1 broke
Pass-rate lift
+45 pp
Tokens
+18%
Agent turns
0%
Response time
-26%
Model gemini-3.6-flashMeasured 8/15/2026v120 of 22 cases passed with skill
Tokens, turns and time describe resource use, not a dollar price.
See it in action
One of 11 test cases the skill fixed. Same model, same prompt, with and without the skill.
1 of 11
Prompt
Reviewer 1 points out that Table 1 reports mean accuracy over 5 runs without standard deviations or confidence intervals. Draft a response formatted with [Priority issue], [Decision dimension], [Draft response], [Evidence anchor], and [Forbidden content removed].
What was checked, without and with the skill
Checked
Without skill
With skill
The Decision dimension is set to statistical validity or reproducibility.
Missed
Passed
The Evidence anchor points to existing variance calculations or supplementary experimental logs.
Missed
Passed
case-16 · gemini-3.6-flash · 8/15/2026 · answers are capped and redacted
What's inside
The instructions your model reads when the skill loads.
SKILL.md65 linessingle file
AISTATS Author Response
Use this after AISTATS reviews are released. Reopen the current OpenReview instructions and
author-discussion policy before drafting because response mechanics are cycle-specific.
Triage
Answer concerns that affect correctness, novelty, statistical validity, clarity,
reproducibility, or fit for AISTATS.
Use existing submitted evidence: paper sections, appendices, supplementary material,
theorem statements, experiments, checklist entries, and code/data descriptions.
Keep the reply anonymous. Do not reveal institution, authorship, grants, private URLs, or
repository ownership.
Treat discussion as clarification, not revision. Do not depend on uploading a revised paper
or new supplement unless current instructions explicitly allow it.
If current rules prohibit links, do not use URLs in the discussion. AISTATS 2026
author-reviewer discussion was text-only and links were not allowed.
Correct factual errors first, then address requests for missing comparisons, uncertainty
estimates, proofs, or hyperparameters.
Drafting pattern
State the decision-critical correction or concession.
Point to exact submitted evidence.
Explain the statistical or theoretical consequence.
Promise a camera-ready wording fix only if it does not add unsupported new claims.
Statistician-reviewer pushback patterns
Pushback
What it signals
AISTATS-ready fix
"Assumption A3 seems strong"
The reviewer traced the proof chain
Point to where A3 is verified, weakened, or shown necessary; never dismiss it as standard without a citation
"The experiments violate the theorem conditions"
Theory-experiment mismatch spotted
Identify which conclusions survive misspecification and cite the robustness simulation that shows it
"Observed rates do not match the bound"
The reviewer compared empirical slopes against theory
Reference the log-log plot, or explain the constant-dominated regime at the tested sample sizes
"Comparison with the classical statistical method is missing"
Statistics-literature gap
Anchor to the appendix comparison, or concede and scope a camera-ready clarification
Response micro-example
Reviewer objection: the minimax claim hides its dependence on dimension d. Reply skeleton: