Audit Section Skill
SkillDev toolsDeep audit of a single section -- style, factual accuracy, citations, logical flow
Available today. Use it from your connected AI after setup.
No other account needed.
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
- Read
.claude/rules/academic-writing.mdfor style rules - Read
.claude/rules/banned-words.mdfor hard/soft bans - Read
.claude/rules/grammar-punctuation.mdfor grammar conventions - Read
.claude/rules/latex-citations.mdfor citation protocol - If the project has
guidance/paper-context.md, read it for correct claims and numbers - 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.mdandacademic-writing.mdfor 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