Whitepaper Audit Agent Skill

Audit a white paper or long-form technical document against a research-grounded best-practices checklist. Two lanes — deterministic script checks (readability, undefined acronyms, structure blocks, broken links) plus an LLM-judge review (overclaims, inconsistent numbers, buried lede, limitations honesty, audience fit). Produces a prioritized P0–P2 findings report by default; applies fixes on explicit request (TDD for any code changes). Use when the user says "audit this white paper", "review my paper against best practices", "check this doc for overclaims", "whitepaper QA", "is this paper ready to publish", or wants prioritized recommendations on a technical document.

13k tokens

context cost

the whole folder, loaded on every use

files

ships runnable scripts

copies elsewhere

how many repositories repackaged it

337

stars on the repo

on the repository, not the skill itself

Install

one command, takes just this skill from the repository

npx skills add https://github.com/glebis/claude-skills --skill whitepaper-audit

What comes with it

48 507 bytes besides the instruction

DESIGN.md

evals/README.md

evals/RESULTS.md

evals/cases/clean-control.md

evals/cases/missing-limitations.md

evals/cases/planted-defects.md

evals/stripped/clean-control.md

evals/stripped/missing-limitations.md

evals/stripped/planted-defects.md

references/audit-prompt.md

references/checklist.md

scripts/check_doc.py

scripts/tests/test_check_doc.py

What it tells the agent to use

found in the instruction text

Task spawns other agents

The instruction itself

10 sections, as written by the author

whitepaper-audit

Audit a markdown white paper in two lanes and produce one merged, prioritized report.

Inputs

Document path (required) — markdown source, not PDF.
Stated audience (ask if not given) — severity of audience-fit/jargon-undefined

depends on it. Default: "technical practitioners, non-academic".

Mode — recommend (default) or fix (only on explicit request).

Workflow

1. Lane 1 — deterministic

python3 scripts/check_doc.py <doc.md> --offline [--target-grade N] [--allow ACRO]

Drop --offline to also check http(s) links (HEAD→GET, timeouts; only *broken* is a

finding). Output: JSON findings, schema in DESIGN.md.

2. Lane 2 — LLM judge

Dispatch a subagent (fresh context — never judge a document you wrote in the same

context) with references/audit-prompt.md, filling {PATH} and {AUDIENCE}, plus the

[judge] criteria from references/checklist.md. The judge returns JSON findings.

Judge calibration rules are binding: verbatim quotes required; no P0 at low confidence;

"needs verification", never "factually wrong".

3. Merge

Dedupe by (location, issue type) keeping both lane attributions; sort P0 → P1 → P2, then

confidence. Cross-reference: a lane-1 broken link that supports a claim (judge decides

materiality) is P1; decorative → P2.

4. Report (default mode)

Write a markdown report: summary verdict, findings table (id, severity, confidence,

location, fix), then details. Recommend; do not edit.

5. Fix mode (only when explicitly requested)

Apply fixes P0-first. Any change to code goes through superpowers

test-driven-development (test first, watch it fail). Prose fixes: edit, then **re-run the

full audit** and report cleared vs remaining findings.

Evals

Before trusting a new/changed judge prompt, run evals/README.md procedure (planted

defects + clean control; pass criteria inside). Lane 1 is covered by

scripts/tests/test_check_doc.py (pytest).

Files

scripts/check_doc.py — lane 1 (stdlib-only; --help for flags)
references/checklist.md — operational criteria, both lanes
references/audit-prompt.md — judge prompt template
evals/ — judge validation cases + pass criteria
DESIGN.md — architecture decisions (v0.2, Codex-audited)

Other skills for the same job

different authors, same section of the catalogue

Hypogenic

by christophacham

×3

Automated LLM-driven hypothesis generation and testing on tabular datasets. Use when you want to systematically explore hypotheses about patterns in empirical data (e.g., deception detection, content analysis). Combines literature insights with data-driven hypothesis testing. For manual hypothesis formulation use hypothesis-generation; for creative ideation use scientific-brainstorming.

Statistical analysis toolkit. Hypothesis tests (t-test, ANOVA, chi-square), regression, correlation, Bayesian stats, power analysis, assumption checks, APA reporting, for academic research.

Automated hypothesis generation and testing using large language models. Use this skill when generating scientific hypotheses from datasets, combining literature insights with empirical data, testing hypotheses against observational data, or conducting systematic hypothesis exploration for research discovery in domains like deception detection, AI content detection, mental health analysis, or other empirical research tasks.

Advanced swarm orchestration patterns for research, development, testing, and complex distributed workflows

9k tokens

Ux Researcher Designer

by ComeOnOliver

×2

UX research and design toolkit for Senior UX Designer/Researcher including data-driven persona generation, journey mapping, usability testing frameworks, and research synthesis. Use for user research, persona creation, journey mapping, and design validation.

Guided statistical analysis for research data - test selection, assumption checking, effect sizes, power analysis, Bayesian alternatives, and APA-formatted reporting. Use whenever a user wants to compare groups, test a hypothesis, analyze experimental or survey data, check statistical assumptions, compute required sample sizes, or write up results - even if they never name a specific test. Covers t-tests, ANOVA, chi-square, correlation, regression, non-parametric and Bayesian methods. For low-level model APIs, see the statsmodels and pymc skills.

29k tokens scripts

Pre Mortem

by phuryn

Run a pre-mortem risk analysis on a PRD or launch plan. Categorizes risks as Tigers (real problems), Paper Tigers (overblown concerns), and Elephants (unspoken worries), then classifies as launch-blocking, fast-follow, or track. Use when preparing for launch, stress-testing a product plan, or identifying what could go wrong.

> Jurisdiction-aware wage/hour and employment Q&A — classification, overtime, meal/rest breaks, leave, final pay — answered for the specific state/country with the controlling rule researched and cited rather than stated from memory. Use when the user asks any employment law question, or says "what's the rule in [state]", "is this exempt", "do we have to pay overtime for", or "can we classify this as".

3k tokens

How to use it

Copy the folder

Take glebis/whitepaper-audit from the repository into ~/.claude/skills for personal use, or into .claude/skills inside a project.

Check the name does not clash

The agent identifies a skill by the name field in its header. Two skills with the same name cannot sit side by side — one of them will be ignored.