Use when drafting an AAAI author response (rebuttal) under the single short character-limited author-feedback window, the no-URL rule, no-new-results guidance, AI-generated-review handling, and the AAAI two-phase review process where Phase-2 papers receive one feedback round before SPC/AC discussion.
SkillScore86Maintained
Automatic
◆Research pack
Last verified 9/2/2026
What it changes
Passed 19 of 22 test cases with the skill, and 7 of 22 without it.
Without skill32%
With skill86%
Checks it passed that the model alone missed
The response draft contains no URLs
The response output contains [Priority issue]
The response output includes the tag [Cut list]
and 17 more in the full report
14 fixed5 passed either way1 still failing2 broke
Pass-rate lift
+55 pp
Tokens
+4%
Agent turns
0%
Response time
-17%
Model gemini-3.6-flashMeasured 9/2/2026v219 of 22 cases passed with skill
Tokens, turns and time describe resource use, not a dollar price.
See it in action
One of 14 test cases the skill fixed. Same model, same prompt, with and without the skill.
1 of 14
Prompt
Reviewer 1 claims that our AAAI paper lacks comparison with Method Z. Method Z is already included under its alternative acronym 'M-Z' in Table 3 on page 6 of our submitted PDF. Help us write a concise 2-sentence rebuttal addressing this reviewer comment.
What was checked, without and with the skill
Checked
Without skill
With skill
The response draft directly cites Table 3 from the submitted manuscript
Passed
Passed
The response output includes the tag [Draft rebuttal]
Missed
Passed
case-12 · gemini-3.6-flash · 9/2/2026 · answers are capped and redacted
What's inside
The instructions your model reads when the skill loads.
SKILL.md64 linessingle file
AAAI Author Response
Use this after AAAI reviews are released. AAAI rebuttal space is scarce: the AAAI-26 FAQ allowed a
single 2500-character response, disallowed URLs, and discouraged new results. AAAI-27 has not yet
published its own rebuttal FAQ — its author feedback window is 2026-10-19 to 10-25 — so treat those
limits as the prior cycle's shape and reopen the current FAQ before drafting.
Triage
Respond to decision-critical factual errors first.
Address human reviews and AI-generated review points when they affect the SPC/AC decision.
Do not spend space on tone, score fairness, or reviewer motivation.
Do not include URLs if current rules forbid them.
Do not introduce new experiments or results. Explain existing evidence, clarify omissions, or
state why requested experiments are not needed.
If supplementary files were missing or corrupted, acknowledge the limitation; AAAI has not
permitted updates at rebuttal time, and the supplementary deadline (2026-07-31 for AAAI-27) closed
months before feedback opens.
What moves an AAAI score
Author feedback at AAAI is read by human reviewers and weighed in SPC/AC discussion, not by the
non-decisional AI review. With one short response and no second round, spend every character on the
concern most likely to flip a borderline decision. Triage each reviewer point into a lane:
Factual error that, if uncorrected, sinks the paper: answer first, cite the submitted anchor.
Misread of an existing result: clarify and point to the table, theorem, or appendix line.
Request for a new experiment: explain why submitted evidence covers it, or scope the claim.
Taste or framing complaint: usually skip; a defensive paragraph rarely moves the score.
Reviewer-pushback patterns and the AAAI fix
"No comparison to method X." Fix: name the submitted table row that includes X, or argue X is out
of scope; do not promise a fresh run that rules forbid.
"Results may not hold beyond one benchmark." Fix: cite the seed or split variation in the
supplement; if absent, narrow the claim instead.
"The AI-generated review flags a missing baseline." Fix: correct it calmly with a submitted
pointer; the AI review is advisory, so a one-line factual rebuttal suffices.