# Cut the Line

> A bait check for unsolicited offers to writers. Paste a message that claims to come from a book club, a reviewer, a literary agent, a publisher, a film or TV company, an award or a marketing service, add the sender's address, and say who wrote first and whether you have already paid anything. The free check runs in the browser: the message as a letter with every bait phrase marked, a tackle tray of seven kinds of bait, the money current (does money flow toward you or away from you?) and a verdict by a fixed rule. One metered reply explains each bait with a word-for-word quote, says which way the money flows and how a genuine offer of that kind usually works, gives three checks to do on your own, drafts a reply you could send and - if you already paid - what to do now.

App: https://cut-the-line.skillsafe.ai/
API tutorial: https://cut-the-line.skillsafe.ai/api.html

## Why it exists

In October 2026 a report on the Hacker News front page described AI-written messages that pose as book clubs, publishers, agents and film scouts, praise a writer's book in convincing detail, and then ask for a "contribution fee" or a "coverage package" - small opening amounts that grow. A genuine offer to a writer usually pays the writer, not the other way round. Cut the Line makes that check quick: what is the bait, and which way does the money go?

## What the free check does

- Marks seven kinds of bait with plain word lists: money asked of you (fees, deposits, packages with a price, invoices, gift cards), flattery ("captivating", "deeply moved", "a rare talent"), a rush (deadlines, spots running out, "within 48 hours"), big-stage promises (film rights, producers, billboards, bestseller lists), off-channel or hush requests ("keep this confidential", "text me"), a generic greeting ("Dear Author") and a free-mail sender address.
- Does not count "no fee" or "we never charge" as an ask. Words about paying you (an advance, royalties, a commission on sales) set the money current toward you; a fee sets it away from you; both can appear.
- The verdict, checked in order: cut the line - they wrote first and the message asks you for money; check the hook first - it asks for money after you contacted them, or shows two or more other kinds of bait; no hook in sight - otherwise.
- Long messages keep their opening and closing with the cut marked. A dropped .eml file fills the sender from its From line.
- The lists are a rough rule of thumb: a hit is a reason to look closer, not proof of a scam, and a clean scan is not proof an offer is real.

## The paid reply

One JSON object: verdict (cut | check | clear), title, summary, lures (one entry per bait the page found, in tray order, each with a word-for-word quote and what it is doing to you, then at most two "other" baits the lists missed), money_flow, real_ones (how a genuine offer of this kind usually works - a general rule of thumb), verify (three checks to do on your own), reply ({subject, body} signed "[Your name]") and if_paid (steps when you already paid, else empty). The page re-checks every reply: verdict against the rule, every found bait explained in order, quotes word for word, no numbers or names that are not in your input, no he/she, no first person outside the draft, no link, address or phone number passed on.

The reply never names a person or organisation in its own words, never repeats a link, address or phone number from the message, and never tells you to use one.

## Exports

Markdown, JSON, a bait sheet (.csv, one row per kind of bait) and the reply as a .txt file. Kept checks go to a logbook in your SkillSafe account (opt-in).

## Source

Built on the SkillSafe skill @claude-office-skills/suspicious-email-analyzer - https://skillsafe.ai/skill/@claude-office-skills/suspicious-email-analyzer (scam and phishing signs in a message: sender, urgency, requests for money or details, impersonation, and what to do next). The reply runs on gpt-terra and is metered per run.