domains / research / paper-drafting / paper-writer
The two skills share the name 'paper-writer' but define fundamentally different procedures. The HKUSTDial skill is a general-purpose, evidence-discipline-focused prose generator that operates on user materials and verified retrieval, explicitly avoiding LaTeX and specific empirical workflows. The brycewang-stanford skill is a specialized, LaTeX-focused workflow for empirical finance/economics papers that requires a PDF input, reads tables/figures, and follows a specific 'write results first' assembly order. They do not reproduce or shorten each other; they are distinct tools for different stages and types of academic writing.
| Repository | Path |
|---|---|
| HKUSTDial/Supervisor-Skills | skills/paper-writer/SKILL.md |
| brycewang-stanford/Auto-Empirical-Research-Skills | skills/67-econfin-workflow-toolkit/paper-writer/SKILL.md |
| Role | Repository | Path |
|---|---|---|
| peer | HKUSTDial/Supervisor-Skills | skills/paper-writer/SKILL.md |
| peer | brycewang-stanford/Auto-Empirical-Research-Skills | skills/67-econfin-workflow-toolkit/paper-writer/SKILL.md |
--- name: paper-writer description: >- Drafts academic paper prose. Use the General mode for STEM/non-STEM prose from notes/ideas with strict evidence tracing. Use the Empirical LaTeX mode for finance/economics papers where a PDF of tables/figures is provided. --- # Paper Writer ## Mode Selection 1. **General Prose Mode**: Use when the user provides notes, ideas, or results and wants publishable prose (paragraph to full manuscript) without a specific LaTeX template or PDF of tables. 2. **Empirical LaTeX Mode**: Use when the user provides a PDF containing tables/figures for an empirical finance/economics paper and wants LaTeX output. ## General Prose Mode (HKUSTDial) ### Hard Rules 1. Every factual claim must trace to user materials, verified retrieval, or field common knowledge (no numbers/names). 2. Zero placeholder tags in output. 3. No invented details. 4. Claim strength matches evidence strength. 5. Distinguish real vs. planned results. ### Workflow 1. **Scope**: Determine granularity (full paper vs. section), paradigm (STEM, humanities, etc.), and mode (Draft vs. Final). 2. **Evidence**: Inventory user materials. Build literature pool via search. Identify gaps. Create Evidence Map for full papers. 3. **Blueprint**: Route by paradigm. Use CER skeleton for non-STEM. 4. **Draft**: Write prose adhering to evidence rules. 5. **Review**: Verify citations and claims. 6. **Deliver**: Clean prose with delivery note. ## Empirical LaTeX Mode (brycewang-stanford) ### Input - PDF path containing tables and figures. ### Workflow 1. **Initialization**: Read PDF, map tables/figures to sections, confirm plan with user, set output path (`main.tex`). 2. **Empirical Results (Write First)**: For each section, ask user for word count, read relevant PDF pages, write LaTeX section, append to `main.tex`. 3. **Research Design**: Write Data/Sample and Variable Construction sections. 4. **Lit Review & Hypotheses**: Write Literature Review and Hypothesis Development. 5. **Introduction & Abstract**: Write Introduction, Title, Abstract, and Institutional Background. 6. **Conclusion**: Write Conclusion. 7. **Assembly**: Finalize `main.tex` with correct document order (Intro, Lit Review, Design, Results, Conclusion). ## Conflicts - **HKUSTDial/Supervisor-Skills**: Focuses on evidence tracing and general prose; does not require PDF or LaTeX. - **brycewang-stanford/Auto-Empirical-Research-Skills**: Requires PDF input, outputs LaTeX, and follows a specific empirical writing order.