Paper excellence

Skill brycewang-stanford/Auto-Empirical-Research-Skills/skills/41-sticerd-eee-sewage-econometrics-check/skills/paper-excellence

🔬 A curated collection of 23,000+ agent skills for empirical research across 8 social science disciplines. | 精选 23,000+ AI Agent 技能库，覆盖8大社会科学学科的实证研究。CoPaper.AI 20分钟完成一篇可复现的规范实证论文，并支持用户上传 Skills。-- Maintained by CoPaper.AI from Stanford REAP.

Install

npx -y skills add brycewang-stanford/Auto-Empirical-Research-Skills --skill paper-excellence

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.

What its author says it does

Copied from the file, not written here

Comprehensive multi-dimensional review of the sewage-house-prices project. Runs econometrics audit, code review, manuscript proofread, and bibliography validation in parallel. Computes a weighted aggregate score. This skill should be used when asked for a "full review", "quality check", "paper excellence", or before submission milestones.

SKILL.md

3.6 KB, as published. Nobody here has run it

Paper Excellence Review

Run a comprehensive quality assessment of the sewage-house-prices project across all dimensions.

Input: $ARGUMENTS — all for full review, or a specific component.

Workflow

Step 1: Identify Targets

docs/overleaf/*.tex — Manuscript sections
scripts/R/09_analysis/ — Analysis scripts
output/tables/ — Generated tables
output/figures/ — Generated figures
docs/overleaf/refs.bib — Bibliography

Step 2: Launch Review Agents (Parallel)

Launch up to 4 agents simultaneously:

Agent 1: Econometrics Audit Review all identification strategies (hedonic, repeat sales, long diff, DiD, upstream/downstream, dry spills). Cross-reference manuscript claims against analysis scripts. Weight: 30%

Agent 2: Code Review Review all scripts in scripts/R/09_analysis/ for code quality, reproducibility, and project convention compliance. Weight: 15%

Agent 3: Manuscript Proofread Review all .tex files for structure, claims-evidence alignment, identification fidelity, writing quality, grammar, and LaTeX compilation. Weight: 35%

Agent 4: Bibliography Validation Cross-reference all citations against refs.bib. Check for missing entries, unused references, and quality issues. Weight: 5%

Step 3: Compute Weighted Aggregate Score

Overall = 0.30 × Econometrics + 0.15 × Code + 0.35 × Paper + 0.05 × Bibliography + 0.15 × Polish

Where Polish is derived from the Proofreader's writing quality subscore.

If components are missing (e.g. no manuscript sections yet), renormalise weights over available components.

Step 4: Present Results

# Paper Excellence Report: Sewage in Our Waters
**Date:** YYYY-MM-DD
**Aggregate Score:** XX/100

## Score Breakdown
| Component | Weight | Score | Issues | Source |
|-----------|--------|-------|--------|--------|
| Econometrics | 30% | XX | N | /econometrics-check |
| Code | 15% | XX | N | /review-r |
| Paper | 35% | XX | N | /proofread |
| Bibliography | 5% | XX | N | /validate-bib |
| Polish | 15% | XX | N | Writing quality subscore |

## Priority Fixes (Top 5)
1. **[CRITICAL]** [Most important issue]
2. **[MAJOR]** [Second priority]
3. ...

## Quality Gate
- Score >= 90: "Ready for submission."
- Score >= 80: "Commit-ready. Address major issues before submission."
- Score < 80: "Blocked. Must fix critical/major issues."

## Full Reports
- Econometrics: output/log/econometrics_check_all.md
- Code: output/log/code_review_all.md
- Proofread: output/log/proofread_report_all.md
- Bibliography: output/log/bib_validation.md

Save to output/log/paper_excellence_[date].md.

Principles

Parallel execution. All agents run simultaneously for efficiency.
Weighted aggregation. Not a simple average — econometrics and paper quality dominate.
Don't double-count. Same issue found by multiple agents counts once in priority list.
One unified report. User sees one priority list, not separate reports.
Proportional gating. Working papers get developmental feedback. Near-final manuscripts get submission-level scrutiny.