◉CLAUDE LABJP
●2.1.295 — Hooks gain onFailure: "block", so a failed or timed-out hook no longer lets the action through●SONNET 4.5 — Retires on the Claude API on 11/30, 52 days left. Migrate to Sonnet 5.5●HAIKU 5.5 — Code written for Haiku 4.5 hits a 400 error on budget_tokens●Q&A — People are asking how to use one connector with two different accounts●API CACHE — Sonnet 5.5 cache reads drop to $0.10/Mtok (Oct 7)●NEW — Before halving always-loaded instruction files: sort what to keep, move, or delete●2.1.295 — Hooks gain onFailure: "block", so a failed or timed-out hook no longer lets the action through●SONNET 4.5 — Retires on the Claude API on 11/30, 52 days left. Migrate to Sonnet 5.5●HAIKU 5.5 — Code written for Haiku 4.5 hits a 400 error on budget_tokens●Q&A — People are asking how to use one connector with two different accounts●API CACHE — Sonnet 5.5 cache reads drop to $0.10/Mtok (Oct 7)●NEW — Before halving always-loaded instruction files: sort what to keep, move, or delete
Articles/Claude.ai
◉ Claude.ai/2026-10-09Beginner

I Kept Rewriting My First Request to Claude Until I Handed It One Page About My Work

If Claude's answers keep missing the mark, the cause may not be your wording. It may be that you never explained your work. Here is how a non-engineer can write a one-page work description today and fold it into the first request.

Claude50prompting4beginner12Projects5work descriptionnon-engineer

One evening I asked Claude to draft a reply to an app store review. The draft came back polite, tidy, and completely unusable.

It opened with a stiff line of thanks that no wallpaper-app user would expect. What I normally write is shorter and a little looser. I rewrote my request three times. The tone still would not land.

My hands stopped over the keyboard, and then it hit me: I had never told Claude what my work actually is.

What was missing before any wording fix

At first I kept polishing phrasing. "More casual." "Shorter." "Less formal." Each fix moved the answer a little closer, and then the next request started from zero again.

The reason is simple. Claude cannot see the context outside our conversation. Who reads this text, where it appears, what I have written before — none of that was in my request.

Think of handing a task to a new colleague. If you only say "write a reply to this review," you will not get what you hoped for. You first explain how your shop works and who the readers are, and only then ask.

I should have done the same with Claude.

A work description: five slots, one page

You do not need a long manual. I use one page with five slots.

SlotWhat to writeMy example
1. The workWhat you make and who it reachesI run wallpaper apps on my own; most users open them on a phone during a commute
2. Where the text goesThe place the writing will appearThe store's review-reply box, with little room for length
3. The readerTheir situation and temperamentSomeone stuck on a bug, or someone who left kind words
4. What I won't doPhrases to avoid, promises I cannot makeNo "fixed in the next update" promises; no boilerplate apology at the start
5. A sample of my writingTwo or three real pieces you wroteA couple of short replies I actually posted

The fifth slot made the biggest difference. Pasting something I had written beat any adjective like "casual." The tone matched much sooner.

Fill it in backwards, starting from the samples

If you go top to bottom, slot one stalls you. Describing your own work in a single sentence is surprisingly hard.

Work backwards instead. Find two or three things you wrote recently — emails, replies, anything. Then recall what you were careful about while writing them, and put that in slot four. Move up to slots three and two, and slot one tends to settle by itself.

I spent about half an hour going top to bottom the first time. Going backwards took roughly ten minutes.

How my first request changed

Before the description, my request was just this:

Please write a reply to the following review.
"The app crashes right after launch. Please fix it."

After:

[About my work]
I run a wallpaper app on my own. The reply appears in the store's review box, so it must be short.
The reader is stuck on a bug. I skip the boilerplate apology and open with a question that clarifies the situation.
I never write "we will definitely fix it in the next update."
Sample: "Sorry for the trouble. Could you tell me your device model and OS version? I'll look into the cause as soon as I know."
 
[This request]
Please write a reply to the following review.
"The app crashes right after launch. Please fix it."

The draft that came back matched the sample's tone almost exactly. I changed a single phrase.

Stop pasting it every time

Copying the page by hand did not last. I skipped it on busy days, and the off-target answers came right back.

So I moved the page into a Claude Project — a place for ongoing work — as its instructions. Conversations started inside a Project carry those instructions along. Screen names and layouts can change with updates, so please check Claude's official help for the current steps.

If you do not use Projects, a note in any memo app works: paste the page at the start of each chat. Where the page lives matters less than making sure it arrives first, every time.

Let it grow, but only where it slips

The first page can be rough. Add lines only where an answer actually slipped.

In my case slot four grew after about two weeks. A reply had come back with an emoji that felt off, so I added one line: no emoji. Adding things in advance only makes the page long and buries the lines that matter.

What to do today

Find two or three short pieces you wrote recently, paste them, and jot down three things you were careful about while writing them. That is already half of the description.

Since I started handing over the page before polishing the request, my rewrites have dropped noticeably.

Share

Thank You for Reading

Claude Lab is ad-free, supported entirely by members like you. We publish practical guides daily with implementation code, benchmarks, and production-ready patterns. If you've found it useful, we'd love to have you on board.

  • ✦Copy-paste ready implementation code
  • ✦New advanced guides published daily
  • ✦$5/mo or $15 for lifetime access
View Membership →

If you found this article helpful, a small tip ($1.50) would mean a lot to us. Your support helps keep this site ad-free and covers server and hosting costs.

Related Articles

◉ Claude.ai2026-06-29
Three Ways to Start with Claude — Browser, Desktop, and Mobile, and How to Use Each
A clear walkthrough for getting started with Claude on the browser, desktop app, and iPhone/Android — plus how to combine all three so your day flows smoothly. Written from the perspective of a solo app developer who lives across all three.
◉ Claude.ai2026-10-03
My Usage Limit Ran Out Before Noon, and the Reason Was Conversation Length, Not Message Count
When the usage-limit banner shows up on Pro or Max sooner than you expected, here is the order I check: the two bars in Settings > Usage, the length of the conversation and its attachments, then the model and effort level. The morning I got cut off before lunch, and the lines I drew afterwards.
◉ Claude.ai2026-09-19
When a proofreading request comes back as a rewrite: keeping the original wording
Ask for proofreading and you often get a polished replacement instead. Fixing the output shape first, splitting edits into three layers, and showing one keep-and-change example pair will hold on to the author's sentence endings and word order.
📚RECOMMENDED BOOKS
Build a Large Language Model (From Scratch)
Sebastian Raschka
LLM Dev
Prompt Engineering for LLMs
Berryman & Ziegler
Prompting
AI Engineering
Chip Huyen
AI Eng
* Contains affiliate links