paper-auditor

paper-auditor is an agent for coding agents from fcakyon/phd-skills. It costs 63 tokens per session (844 once invoked), scanned A, original, MIT.

An agent that checks whether a research paper agrees with its source code, data, and experiment results. It examines LaTeX files, citations, result files, configurations, logs, and related code.

In plain words
What is it for?
Use it to verify paper numbers, trace claims to code or results, check that methods match the implementation, and produce a prioritised consistency report.
Why use it?
It helps find unsupported claims, incorrect numbers, and differences between the described method and the implementation. It reports issues with their locations and suggested fixes.

Agent

Part of the phd-skills plugin — 12 skills, 6 commands, 2 agents, 4 hooks shipped together

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add agents/fcakyon/phd-skills/paper-auditor
Clone the repo
git clone --depth 1 https://github.com/fcakyon/phd-skills

Or install phd-skills, the plugin that ships this one along with the rest of its 12 skills, 6 commands, 2 agents, 4 hooks.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for paper-auditor

README.md
[![agentmods](https://agentmods.dev/badge/agents/fcakyon/phd-skills/paper-auditor.svg)](https://agentmods.dev/agents/fcakyon/phd-skills/paper-auditor)
Your own site
<a href="https://agentmods.dev/agents/fcakyon/phd-skills/paper-auditor"><img src="https://agentmods.dev/badge/agents/fcakyon/phd-skills/paper-auditor.svg" alt="Measured on agentmods" height="20"></a>
Per session 63 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 844 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00063 $0.00844
Opus 5 $0.00032 $0.00422
Sonnet 5 $0.00013 $0.00169
Haiku 4.5 $0.00006 $0.00084

Measured 5d ago against content hash 3bd7bafec4ca, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

paper-auditor scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 5d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

plugin/agents/paper-auditor.md · 119 lines

How it starts

The opening of the file, as written. The whole thing — 119 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Paper Auditor Agent

You are an autonomous agent that audits a research paper for consistency with its codebase and experimental results. You work in an isolated worktree to avoid affecting the user's working directory.

Your Mission

Systematically verify that every claim in the paper is supported by code, data, or experimental results. Produce a prioritized list of issues with specific fixes.

Audit Protocol

Phase 1: Discovery

  1. Find the paper's .tex files (Glob for **/*.tex)
  2. Identify the main document and its structure
  3. Find .bib files for citation data
  4. Locate result files, configs, evaluation scripts, and training logs
  5. Identify the key code modules referenced by the paper's methods

Phase 2: Numerical Audit

For every number in the paper:

  1. Extract the number and its context (section, sentence)
  2. Search the codebase for its source (result files, configs, logs)
  3. Verify the value matches exactly
  4. Note rounding conventions and check consistency

Record in a table:

| Claim | .tex Location | Source | Source Value | Match? |

Phase 3: Method-Code Alignment

For each method described in the paper:

  1. Find the implementing code (function, class, module)
  2. Compare algorithm steps in paper vs code flow
  3. Check hyperparameters in text match defaults/configs
  4. Verify architecture descriptions match model definitions
  5. Confirm loss function equations match code

Phase 4: Terminology Consistency

  1. Extract defined terms from the methods section
  2. Search all sections for each term
  3. Flag: same concept different names, same name different meanings

Phase 5: Citation Spot-Check

For the 5 most important citations:

  1. Verify author names and venue against DBLP via web search
  2. Check any specific numbers attributed to cited papers
  3. Flag unverifiable claims

Phase 6: Evaluation Integrity

  1. Read evaluation scripts
  2. Check for data leakage between splits
  3. Verify metric computation (aggregation method, edge cases)
  4. Run evaluation if possible and compare output to paper values

Read the full file on GitHub · 119 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 5d ago First seen · 119 lines · 63 tokens per session scan A 3bd7bafec4ca

Subscribe to this mod's changes

paper-auditor is an agent published in the GitHub repository fcakyon/phd-skills (385 stars, last pushed 2mo ago), licensed MIT. It adds 63 tokens to every session and 844 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Related

Other agents, from other repositories

geometrist

Classical and computational geometry specialist for ruler-and-compass constructions, Euclidean proofs, trigonometric problem solving, and geometric transformations.

pjt222/agent-almanac · 30 tokens

metadata-extractor

Extracts paper metadata (authors, date, venue, fields, DOI/arxiv ID) and a paper-quality assessment (credibility, experimental rigor, reproducibility) from a paper's plain text. Invoked alongside lite-drafter and finding-extractor during /paperloom:ingest.

trapoom555/claude-paperloom · 63 tokens

source-ingester

Специализированный агент по ингесту внешних источников (статья, книга, видео, доклад, транскрипт, конспект лекции, экспорт цитат) в персональную базу знаний Obsidian. Делегируй сюда, когда пользователь хочет «обработать статью», «положить книгу в базу», «выжать знания из видео», «обработай конспект лекции», «разбери…

jtprogru/bear-skills · 181 tokens

clinical-modeler

Use this agent when the user needs to read, write, review, or edit local archetype (.adl), template (.oet, Archetype Designer .t.json, .opt), or composition files in the workspace. It writes only to the local workspace, and can perform read-only MCP lookups (terminology resolution, RM/AM type specs, guides, and…

Cadasto/openehr-assistant-plugin · 407 tokens

note-doctor

Специализированный агент для работы с одной конкретной заметкой в Obsidian — обогащение frontmatter (aliases, up, down, other), критика, поиск противоречий и слабых мест, peer review. Делегируй сюда, когда пользователь говорит про конкретную заметку: «обогати эту заметку», «дозаполни frontmatter», «добавь aliases»…

jtprogru/bear-skills · 194 tokens

inbox-triager

Специализированный агент по обработке inbox (00. Входящие/) в персональной базе знаний Obsidian. Делегируй сюда, когда пользователь хочет «разобрать входящие», «прочистить inbox», «разложить.

jtprogru/bear-skills · 58 tokens