Use when a model — a full theory paper's argument or the structural/conceptual scaffolding behind an empirical AER: Insights short-format manuscript — needs to be made minimal, so it carries the single insight in a few pages. Keeps only what the insight requires; it does not develop a general framework.
SkillScore95Maintained
Automatic
◆Research pack
Last verified 8/15/2026
What it changes
Passed 21 of 22 test cases with the skill, and 12 of 22 without it.
Without skill55%
With skill95%
Checks it passed that the model alone missed
Specifies aeri-robustness as the next step.
Includes the tag 【Case】 set to theory-is-the-paper.
Includes the tag 【Next step】 set to aeri-robustness.
and 9 more in the full report
9 fixed12 passed either way1 still failing0 broke
Pass-rate lift
+41 pp
Tokens
+4%
Agent turns
0%
Response time
-49%
Model gemini-3.6-flashMeasured 8/15/2026v121 of 22 cases passed with skill
Tokens, turns and time describe resource use, not a dollar price.
See it in action
One of 9 test cases the skill fixed. Same model, same prompt, with and without the skill.
1 of 9
Prompt
We are submitting an empirical paper where a dynamic labor supply model motivates our empirical identification. We want to know what follow-up skill or module should be consulted immediately after applying the single-insight theoretical modeling discipline.
What was checked, without and with the skill
Checked
Without skill
With skill
Specifies aeri-robustness as the next step.
Missed
Passed
case-18 · 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.md80 linessingle file
Theory / Model — Minimal to Carry One Insight (aeri-theory-model)
When to trigger
A theory paper's single result (a theorem, an impossibility, a sharp comparative static) needs a tight model
An empirical paper carries a structural or conceptual model that has grown beyond what the result needs
The model section is eating the word budget that the result and identification need
A referee might ask "why all this machinery for one point?"
The AER: Insights modeling discipline
AER: Insights rewards a single, surprising, clean theoretical or modeling insight delivered with the smallest model that produces it. The question is never "is the framework general?" but "what is the most economical model whose one result is important and surprising?" Generality, extensions, and robustness of the modeling go to the Supplemental Appendix; the main text states the assumptions, the one key result, and the intuition. A theory paper here is short and pointed, the opposite of a foundational framework paper for AER or a field journal.
Two cases
Case 1: The paper IS the theory (a theory short paper)
One headline result. A single theorem, impossibility, or characterization that overturns or sharply refines a believed claim. State it early, prove it economically.
Minimal assumptions. Strip to the assumptions the result actually needs; flag which are substantive vs. technical.
Intuition before proof. Give the one-paragraph mechanism that makes the result memorable; full proofs to the appendix if long.
Surprise stated. Name the prior intuition the result violates — that is why it is an AER: Insights paper, not a field-theory note.
Case 2: The model SUPPORTS an empirical insight
Justify the model's job. It exists to deliver one object (an identifying restriction, a counterfactual, a structural parameter the data identify). Cut anything not in service of that.
Minimal structure for identification. Tie the key parameter to a data moment (aeri-identification Branch C); keep solution/estimation detail in the appendix.
One counterfactual, if any. If the structure delivers a counterfactual, report the single one that matters with its uncertainty; alternatives to the appendix.
What stays in vs. goes to the appendix
In the main text (few words)
Supplemental Appendix
Setup + key assumptions
Proofs (if long), lemmas, regularity conditions
The one headline result + intuition
Generalizations, extensions, alternative specifications
The one identifying restriction / counterfactual
Robustness of the model, additional counterfactuals
Word budget for the model section
The model competes with identification, result, and discussion for the same ≤7,000-word (minus exhibits)
budget (aeri-writing-style). A useful rule: the in-text model should be
sketchable on a single page — assumptions, the one result, the intuition. If the derivation cannot be
sketched that compactly, the full version belongs in the Supplemental Appendix with only the statement and