---

mode: empirical
discipline: interdisciplinary
title: "<Working title of the study>"
---

# Research brief — <working title>

This file is the complete brief for one autonomous AutoScholar research run. Write it for a careful researcher who has never heard of the platform: everything they would need to know to carry out the study is here. Replace every <placeholder>.

## 1. Research objective and scope

- **Question.** <One precise question, phrased so that it can be answered negatively.>
- **Why it matters.** <Two or three sentences on the scientific or practical stake.>
- **Scope.** <What is in and out: population, period, geography, model class, text corpus…>
- **Non-goals.** <What the study deliberately does not attempt.>

## 2. Domain, mode and methodology

- **Discipline(s):** <economics | finance | mathematics | computer science | medicine | … >
- **Primary mode:** <empirical | computational | theoretical | review | experimental | humanities>
- **Secondary modes (optional):** <…>
- **Methodology.** <Name the method and why it fits the question: e.g., panel regression with two-way fixed effects; proof by induction with a counterexample search; PRISMA-style systematic review; benchmark with 5 seeds and ablations.>

## 3. Proposed contribution and novelty assessment

- **Contribution.** <What new knowledge should exist after the run.>
- **Novelty check.** <Which prior work to compare against; what would make the result "already known".> State novelty in qualified terms ("to our knowledge").

## 4. Literature and source rules

- Search at least <2–3> indexes (OpenAlex, Crossref, arXiv, Semantic Scholar, PubMed as fits the field). Target <10–30> verified sources.
- Cite only sources added through the platform's verification tool; prefer DOIs; read full texts where open access.
- Required reading: <list any must-cite references with DOI/arXiv id>.
- Record a claim-to-source map for every background claim used in the introduction and discussion.

## 5. Evidence: data, simulations, proofs or texts

- **Data sources (empirical):** <official APIs/repositories, exact tables/series, period, license>. Document provenance, license, download commands, hashes.
- **Simulation (if any):** <design, seeds, what it stands for>. Label synthetic output as synthetic.
- **Proof/formal objects (theoretical):** <definitions, assumptions, candidate propositions>.
- **Corpus (review/humanities):** <sources, editions, inclusion criteria>.

## 6. Procedure

1. <Step-by-step protocol: cleaning, estimation/derivation/screening, outputs.>
2. <Pre-registered specifications or proof obligations to run first.>
3. <Secondary analyses allowed only if labelled exploratory.>

## 7. Baselines, robustness and falsification

- **Baselines/alternatives:** <…>
- **Robustness:** <alternative specifications, samples, hyper-parameters, relaxed assumptions, counterexample search>.
- **What would falsify the claim:** <explicit conditions under which the paper must report a null or negative result>.

## 8. Reproducibility and artifacts

- All code under `code/` with a pinned environment and `code/reproduce.sh` regenerating `results/metrics.json`, tables and figures from raw inputs.
- All reported numbers registered with the platform's `record_reported_value` tool.
- Data manifest with licenses and SHA-256; experiment log with commands, seeds, exit codes.

## 9. Manuscript structure and LaTeX standards

- AutoScholar class (`\documentclass{autoscholar}`), natbib author-year, booktabs tables, vector figures.
- Sections: abstract (150–300 words), introduction, related work, method/theory, data (if any), results, robustness, discussion incl. competing explanations, limitations, conclusion, references, appendices (proofs, protocol, reproducibility).
- Target length: <8–20 pages>; depth over length.

## 10. Critical review, revision and quality expectations

- The independent reviewer should focus on: <identification/validity threats, proof gaps, screening bias, …>.
- Revise substantively; document a point-by-point response.
- Hard requirements: verified citations, traceable numbers, honest limitations, clean build.

## 11. Limitations, ethics and legal constraints

- <Known limitations the paper must state.>
- <Ethical/legal constraints: no human-subject work, data licenses, no copyrighted full-text republication.>

## 12. Acceptance criteria

The run is successful only if: <concrete, checkable criteria — e.g., "the main specification and two robustness checks are estimated on the full sample; the paper reports effect sizes with confidence intervals; the proof of Proposition 1 is complete and checked symbolically; at least 15 verified sources are cited">.