ai-asset-pricing: Skill for Claude Code

Full Paper Audit Skill

Comprehensive audit of the entire paper for style compliance, factual consistency, citation correctness, and cross-section coherence.

Examples

• /full-paper-audit -- run complete audit
• /full-paper-audit --focus style -- style-only pass
• /full-paper-audit --focus citations -- citation-only pass

Workflow

Step 1: Load All Context

1. Read .claude/rules/academic-writing.md
2. Read .claude/rules/banned-words.md
3. Read guidance/paper-context.md (if it exists)
4. Read .claude/rules/latex-citations.md (if it exists)
5. Read main.tex in full

Step 2: Section-by-Section Audit

Discover sections by scanning for %% BEGIN: / %% END: markers in main.tex. Run /audit-section on each section sequentially.

Step 2b: Math Audit

For sections containing formal environments (any section with \begin{proposition}, \begin{theorem}, \begin{lemma}, or \begin{proof}), run /audit-math. Include SEVERITY SUMMARY and TOP PRIORITY FIXES in the master report.

Step 2c: Editorial Artifact Scan

Before proceeding to cross-section checks, grep the full manuscript for submission-blocking editorial artifacts in active prose (not LaTeX % comments):

• [HUMAN EDIT, TODO, FIXME, XXX, [TBD], [PLACEHOLDER], [INSERT
• Parenthetical editing notes: (change to, (should be, (need to, (fix this), (update this)
• Missing-reference markers: [??], [?], [cite], [ref] Any hit is Critical and goes to the top of the priority fixes list.

Step 3: Cross-Section Consistency

Check that the SAME numbers are used consistently everywhere:

• Do quantitative claims in the introduction match those in the results sections?
• Does the conclusion match the findings reported in the body?
• Are terminology choices consistent across all sections?
• Are counts (sample size, number of variables, etc.) consistent everywhere they appear?

If guidance/paper-context.md exists, cross-reference all claims against its canonical values.