Use when assembling the data, code, and documentation package for an American Economic Journal: Macroeconomics (AEJ: Macro) manuscript to pass the AEA Data and Code Availability Policy and the AEA Data Editor's pre-publication reproducibility check. Covers macro specifics (simulation/calibration code, restricted-access data); it does not write the analysis itself.
SkillScore91Maintained
Automatic
◆Research pack
Last verified 8/15/2026
What it changes
Passed 19 of 21 test cases with the skill, and 16 of 21 without it.
Without skill77%
With skill91%
Checks it passed that the model alone missed
Output contains the bracketed tag 【Long runs】
Output contains the bracketed tag 【Repository】
Output contains the bracketed tag 【Code scope】
and 4 more in the full report
5 fixed15 passed either way0 still failing2 broke
Pass-rate lift
+14 pp
Tokens
+19%
Agent turns
0%
Response time
-28%
Model gemini-3.6-flashMeasured 8/15/2026v120 of 22 cases passed with skill · 21 comparable
Tokens, turns and time describe resource use, not a dollar price.
See it in action
One of 4 test cases the skill fixed. Same model, same prompt, with and without the skill.
1 of 4
Prompt
We just received conditional acceptance at AEJ: Macro for our heterogeneous-agent DSGE paper, and we need to assemble our replication code and data before submitting to the AEA Data Editor. Our materials include Dynare code, custom Julia solver scripts, and confidential administrative data from a national central bank. Please perform an audit of our packaging plan against journal reproducibility standards. Provide a structured verification breakdown covering openICPSR readiness, master script execution scope, full code inclusion across solvers and simulations, toolchain and random seed configuration, handling of restricted data, and management of heavy computational runs, followed by the recommended next step.
What was checked, without and with the skill
Checked
Without skill
With skill
Output contains the bracketed tag 【Repository】
Missed
Passed
Output specifies aejmac-referee-strategy as the next step
Missed
Passed
case-01 · 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.md66 linessingle file
Replication Package (aejmac-replication-package)
When to trigger
A paper is heading toward conditional acceptance and the AEA Data Editor check is next
You have simulation/calibration code but have never packaged it for a reviewer to run
The data include a restricted/proprietary source (confidential micro data, licensed series)
You want to build the package as you go rather than scrambling at acceptance
The AEA reproducibility regime (verified 2026-06; re-confirm on the official AEA pages)
Governed by the AEA Data and Code Availability Policy. Conditionally accepted papers undergo a review by the AEA Data Editor (Lars Vilhuber) before publication, including reproducibility checks and verification of the information provided.
Deposit in the AEA Data and Code Repository on openICPSR (use is strongly encouraged; other trusted repositories may be allowed with Data Editor approval). Materials are posted with the article.
Code scope is broad — and this is the macro-critical point: the policy covers data cleaning and "estimation, simulation, model solution, and visualization" code. For AEJ: Macro, the DSGE/HANK solver, the calibration/estimation routines, and the simulation code must all be in the package, not only the regression scripts.
Restricted-access data: exceptions exist for confidential / copyrighted / agreement-restricted data; authors must preserve materials 5+ years, provide reasonable replication assistance, make the code public even when the data cannot be, and disclose data sources. State any such request to the Data Editor.
Field experiments must be registered in the AEA RCT Registry.
One master script (run_all) that regenerates every table and figure from raw inputs in order, including the model solution and simulation steps.
A README following the AEA template: data sources and access, software + versions, hardware, expected runtime, and a map from each exhibit to the script that makes it.
Macro-specific reproducibility
Pin the toolchain: Stata version + ssc/net package versions; R renv.lock; Python requirements.txt/conda env; Julia Project.toml/Manifest.toml; Dynare version for DSGE.
Seeds set and reported for every simulation, bootstrap, and randomization step.
Long-running computations (global solutions, large HANK simulations, MCMC): provide a way to verify without a supercomputer — ship intermediate/cached outputs and a reduced-scale switch, and document expected full runtime.
Numerical accuracy artifacts: include the diagnostics (Euler errors, grid checks) so the Data Editor can confirm the solution, not just rerun it.
Data documentation
For each source: provider, exact extract/vintage, access date, license, and whether it is public or restricted.
Real-time vs. revised macro vintages (e.g., ALFRED): document which you used.
Restricted data: a clear access path and a public code subset that runs on synthetic/sample data where possible.
Checklist
openICPSR (AEA Data and Code Repository) deposit planned; README on the AEA template
Master run_all regenerates every exhibit incl. model solution + simulation
Simulation, calibration/estimation, and solver code all included (not just regressions)
Use when assembling the proof appendix and any code/data deposit for an American Economic Journal: Microeconomics (AEJ: Micro) manuscript under the AEA Data and Code Availability Policy. Covers proof appendices for pure theory plus numerical/structural/experimental code; it does not run your estimation.