Audit Section Skill

SkillDev tools

Deep audit of a single section -- style, factual accuracy, citations, logical flow

Available today. Use it from your connected AI after setup.

Connect ahel once, and every AI you use reads what you have installed.

Then ask your AI: use the Audit Section Skill skill

What this skill tells your AI

The instructions your AI receives, as published by alexander-m-dickerson/ai-asset-pricing in .claude/skills/audit-section/SKILL.md and read by ahel’s review.

Comprehensive single-section audit combining style checking, factual verification, citation auditing, and logical flow analysis.

Examples

  • /audit-section introduction -- deep audit of the introduction
  • /audit-section data -- audit the data section
  • /audit-section results -- audit the results section

Workflow

Step 1: Load Context

  1. Read .claude/rules/academic-writing.md for style rules
  2. Read .claude/rules/banned-words.md for hard/soft bans
  3. Read .claude/rules/grammar-punctuation.md for grammar conventions
  4. Read .claude/rules/latex-citations.md for citation protocol
  5. If the project has guidance/paper-context.md, read it for correct claims and numbers
  6. Read the target section from main.tex (use /extract-section)

Step 2: Style Audit

Run the full /style-check analysis:

  • Banned words, throat-clearing, passive voice, superlatives, vague claims, self-praise
  • Structural AI tells: em-dashes, AI-marker words (per Kobak/Liang), naked "this", adverb openers, "Together, these results...", soft-ban counts
  • Hedge words & previewing: somewhat/quite/very/arguably/perhaps (Nikolov); "as we show below"/"Recall from" (Cochrane); nominalizations (Williams)
  • See banned-words.md and academic-writing.md for the full current lists

Step 3: Factual Accuracy

If the project has guidance/paper-context.md, cross-reference every quantitative claim:

  • Do numerical claims match the paper's canonical values?
  • Do table references match actual table content?
  • Flag any inconsistency between text claims and tables/figures

If no paper-context file exists, flag claims that cannot be verified.

Step 4: Citation Audit

For each citation in the section:

  • Verify key exists in .bib
  • Check citation supports the claim being made (not just existence but relevance)
  • Flag citations used out of context

Step 5: Logical Flow

  • Does the section follow a logical progression?
  • Are transitions between paragraphs smooth?
  • Is there redundancy (same point made twice)?
  • Does the opening paragraph set up what follows?
  • Does the section deliver on its implicit promise?

Step 6: Economic Reasoning

  • Are economic arguments sound?
  • Are mechanisms explained correctly and consistently with the paper's framework?
  • Do the results follow from the methodology described?

Output

SECTION AUDIT: [section name]
================================

STYLE ISSUES: N total (M critical, K suggestions)
[categorized list]

FACTUAL ACCURACY:
- [list of verified/flagged claims with specific numbers]

CITATION AUDIT:
- [status of each citation in section]

LOGICAL FLOW:
- [structural observations and suggestions]

ECONOMIC REASONING:
- [any issues with mechanism descriptions]

PRIORITY FIXES:
1. [highest priority issue]
2. [second priority]
3. [etc.]

SUGGESTED REWRITES:
[specific rewrite suggestions for the worst passages]

Signals

GitHub stars
59
Forks
10
Last commit
Apr 2026
Advanced
Catalog kind
skill
Gateway key
audit-section
Source
github.com/alexander-m-dickerson/ai-asset-pricing