paper2pr: Agent for Claude Code

.claude/agents/domain-reviewer.md

domain-reviewer is an agent for Claude Code from alohays/paper2pr. It costs 57 tokens per session (1,496 once invoked), scanned A, original, MIT.

A fact-checking agent for presentation slides. It compares dates, numbers, names, autonomy labels, and other claims with the sources declared by the deck.

In plain words
What is it for?
Use it after drafting slides or before presenting to verify factual claims and check the deck's forbidden-term list.
Why use it?
It identifies unsupported or inaccurate claims before the deck is presented.

Agent for Claude Code

Written for Claude Code: installed under .claude/. Also seen: model in frontmatter.

This is alohays/paper2pr's own configuration. It tells Claude Code how to work on paper2pr itself, so it is not a mod to install elsewhere. Copy it as a starting point and replace the rules that are about this project. Everything paper2pr configures →

Needs its repository: it runs a file that does not travel with it, so clone the repository first. The line is python3 scripts/deckprofile.py <Deck> # resolved profile JSON.

Reuse

Borrowing it

Nothing to install: this file belongs to alohays/paper2pr. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file
curl -O https://raw.githubusercontent.com/alohays/paper2pr/main/.claude/agents/domain-reviewer.md
Clone the repo
git clone --depth 1 https://github.com/alohays/paper2pr

Made for: Claude Code.

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 domain-reviewer

README.md
[![agentmods](https://agentmods.dev/badge/agents/alohays/paper2pr/domain-reviewer.svg)](https://agentmods.dev/agents/alohays/paper2pr/domain-reviewer)
Your own site
<a href="https://agentmods.dev/agents/alohays/paper2pr/domain-reviewer"><img src="https://agentmods.dev/badge/agents/alohays/paper2pr/domain-reviewer.svg" alt="Measured on agentmods" height="20"></a>
Per session 57 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,496 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
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 5.1 $0.00057 $0.01496
Opus 5 $0.00028 $0.00748
Sonnet 5 $0.00011 $0.00299
Haiku 4.5 $0.00006 $0.00150

Measured 8d ago against content hash b5498acc1fc2, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-08, from the pricing page.

Security

Grade A, and why

domain-reviewer 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 8d 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.

.claude/agents/domain-reviewer.md · 140 lines

How it starts

The opening of the file, as written. The whole thing — 140 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Fact check

You are the fact-check agent. (The frontmatter name stays domain-reviewer so existing references hold; the job is fact checking.) Your job is NOT presentation quality - that's other agents. Your job is whether the factual content of the slides survives comparison with the deck's own sources.

Step 0: load the deck's premises (always, before reading a slide)

python3 scripts/deckprofile.py <Deck>                          # resolved profile JSON
cat "$(python3 scripts/deckpath.py <Deck> --field config)"     # <deck>.deck.yml

From the profile JSON take sources (the list of paths and URLs the deck declares under sources: in deck.yml) and forbidden_file (the path of <deck>.forbidden.txt, or null). The audience comes from deck.yml's audience block (fall back to the audience: section of .claude/rules/slide-profiles/<profile>.yml); state it in the report header - it decides how much simplification is fair.

If sources is empty, write a one-paragraph report saying the deck declares no sources so no claim on it can be verified, list the 5-10 most load-bearing factual claims that would need one, and stop. Do not fact-check against your own memory of the field; memory is not a source.

Step 1: read every source

Read each entry of sources end to end before reading the slides:

  • A path (absolute, or relative to the repo root) is read with Read. For W02 and other lecture decks these are vault research notes; read them as the source of record even though they live outside this repo.
  • A URL is fetched with WebFetch.
  • A source that cannot be read is reported as such, by name, and the claims that depended on it are marked unverifiable - never silently skipped.

Read forbidden_file too, when it exists (one term per line, # comments).

Step 2: compare the slides against the sources

Walk every slide of the qmd and check, claim by claim:

Dates

  • Every year, month, or "in 20XX" on a slide appears in a source with the same value. Timeline slides (.timeline items) are checked entry by entry.

Read the full file on GitHub · 140 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. 8d ago First seen · 140 lines · 57 tokens per session scan A b5498acc1fc2

Subscribe to this mod's changes

domain-reviewer is an agent published in the GitHub repository alohays/paper2pr (6 stars, last pushed 6d ago), licensed MIT. It adds 57 tokens to every session and 1,496 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 2026-08-31.