Claude Code skill family for writing, polishing, and submitting statistics papers to Big Four journals (JASA, AoS, JRSS-B, Biometrika) and similar venues (AOAS, JASA ACS, EJS, Bernoulli,...
概览
A family of Claude Code skills for writing, polishing, and submitting statistics manuscripts to the Big Four statistics journals (JASA, Annals of Statistics, JRSS-B, Biometrika) and similar venues (AOAS, EJS, Bernoulli, Statistica Sinica, Biostatistics, JCGS, JASA ACS), plus ML theory venues (COLT, ALT). The skills support three paper types with distinct section structure and theory weight: - (AoS, Bernoulli, JRSS-B, EJS, COLT, ALT, MSL) - (JASA T&M, JRSS-B, Biometrika, Statistica Sinica) - (AOAS, JASA ACS, Biostatistics, Statistics in Medicine, JCGS, JABES) Located in stat-shared-references/. These are reusable across the four skills. Mechanical checks live in tested Python (stat-shared-references/scripts/, stdlib only) so the skill bodies stay focused on judgment. Every script separates findings, which affect the exit code, from ones, which never do.
README
以下为原文 README 节选，完整内容请查看原文。
stat-writing-skills
A family of Claude Code skills for writing, polishing, and submitting statistics manuscripts to the Big Four statistics journals (JASA, Annals of Statistics, JRSS-B, Biometrika) and similar venues (AOAS, EJS, Bernoulli, Statistica Sinica, Biostatistics, JCGS, JASA ACS), plus ML theory venues (COLT, ALT).
The skills support three paper types with distinct section structure and theory weight:
- Theory papers (AoS, Bernoulli, JRSS-B, EJS, COLT, ALT, MSL)
- Methodology papers (JASA T&M, JRSS-B, Biometrika, Statistica Sinica)
- Application papers (AOAS, JASA ACS, Biostatistics, Statistics in Medicine, JCGS, JABES)
Skills
| Skill | Purpose |
|---|---|
stat-paper-plan |
Build a structured outline from a narrative report or research notes. Produces PAPER_PLAN.md, PRIOR_WORK_MATRIX.md, TECHNICAL_RISK_REGISTER.md |
stat-paper-write |
Draft section-by-section LaTeX from the plan. Builds and audits CLAIM_SUPPORT_MAP.md. Handles main/supplement separation per venue. |
stat-paper-writing |
Full pipeline orchestrator: plan → figures → write → compile → improvement loop with Claude internal + Codex external review |
stat-polishing |
Polish existing prose to Big Four standards. Positioning audit, technical claim strength audit, then style discipline pass. Includes page-fit micro-edit and three-reader pre-submission test modes. |
stat-mock-review |
Single-pass AE-style pre-submission mock review. Produces MOCK_REVIEW.md with synopsis, fatal / major / minor concerns, venue-fit risk, likely initial editorial action (verb, not number), and a prioritized rescue plan. |
Shared References
Located in stat-shared-references/. These are reusable across the four skills.
| File | What it covers |
|---|---|
stat-writing-principles.md |
Narrative arc, paper types, abstract formulas, introduction structure |
stat-style-discipline.md |
Punctuation discipline (em-dash, colon, semicolon), emphasis-formatting discipline (no bold/italic in prose, venue table), AI-template removal, AI-vocab watchlist, Biometrika-style house bans (note that, is given by, homemade acronyms), COPSS-style scholar voice, paragraph and bullet discipline |
stat-figure-design.md |
No titles in figures, caption discipline with sentence-style + per-venue table, statistical-question-driven figure-type selection, Big Four guardrails (no violins without sample support, no broken axes, no dual-y, no inset legends in Biometrika, no 3D, no pie, no splash framework figures), legend placement, sizing for journal columns, multi-panel rules |
stat-theory-writing.md |
Theorem statement patterns, assumption blocks, proof sketches, rate comparison tables, minimax lower bound arguments |
推荐工具
换一个关键词，或者移除筛选条件。
安装
npx skillfish add gyf9712/stat-writing-skills