Shelf
See how it works · Writer's Block Killer

How Writer's Block Killer works

Your expertise, faster.

Here's the full flow, end to end. A founder writes the form: what they're stuck on, who's reading it, samples of how they write. The AI returns a polished draft in their writing style, not in default AI tone.

The actual tool runs locally and works the way you'll see below. This page is the preview.

Step 01

The tool opens with three empty form sections

Double-click Writers-Block-Killer.html from your downloads. It opens in any browser. No install, no account, no API key.

file:///Writers-Block-Killer.html
Shelf HQ v1.3
Section 1 of 3

What are you writing?

Describe it however you want. The more detail you give, the better the draft you get back.

Describe your writing task...
Section 2 of 3

Tell me a bit more

All optional. Skip anything that doesn't apply to you.

Optional context...
Section 3 of 3

Show me how you write

Paste 2-3 short samples. This is how the AI learns to sound like you instead of like an AI.

Paste your first writing sample here...
Pastes a ready-to-go prompt for ChatGPT, Claude, or Gemini.
Step 02

Describe what you're writing and where you're stuck

The buyer here is a founder writing a launch-announcement blog post about an infrastructure rebuild. They've been trying to write it all morning. It keeps coming out like a press release. They want it to sound like themselves.

Section 1 of 3

What are you writing?

Describe it however you want. The more detail you give, the better the draft you get back.

A blog post announcing our auto-scaling rebuild for our existing customer base. Around 500 words. Lead with the "why," not the "what." We had a couple of high-profile downtime incidents in Q1 that we already wrote post-mortems for; this is the followup that says we fixed the root cause and ships the rebuild today.
Step 03

Add context. Show the AI how you write.

Section 2 sharpens audience and tone. Section 3 is the writing-style match step. The buyer pastes 2 short writing samples in the same conversational register they want the draft to land in. The AI extracts pattern from the samples (sentence rhythm, opener style, contraction usage) without copying the words.

Section 2 of 3

Tell me a bit more

Audience
Our customers. Mostly engineering leaders at mid-market SaaS.
Tone
Friendly-direct. Plain-language. Not pitchy.
Length target
~500 words
Urgency
Posting Friday
What's the hard part?
I keep writing it like a press release. Generic, smooth, forgettable. I want it to sound like I'm explaining this to a friend over coffee, not announcing it to a room of investors.
Section 3 of 3

Show me how you write

Paste 2-3 short samples. This is how the AI learns to sound like you instead of like an AI.

Sample 1 (required)
Most companies I've worked with have one piece of infrastructure that holds them hostage. For us it was the queue. We rebuilt it last quarter. The first week after launch we got eight customer tickets that all said the same thing: "whatever you did, do it again." That's the bar I hold every roadmap decision to now.
Sample 2 (recommended)
Spent three hours yesterday in a sprint planning call where someone said "we should productize the consultant tier." Eight people nodded. Nobody could name a customer who'd pay for it. We didn't ship a productized tier. We shipped a Friday call instead. Talked to actual customers. Found three real problems. None of them was the productized tier.
Humanize mode on. The AI scrubs em-dash clusters, hedging words, AI-tell transitions, and corporate-warm filler. Your draft reads as a careful human writer's work.
Step 04

Click Build. The full prompt assembles.

One click. The tool stitches your task, your context, your samples, and the 4-phase methodology (Diagnose, Outline, Draft, Polish) into a paste-ready prompt. Copy with one button. Open ChatGPT, Claude, or Gemini directly with the next.

Your prompt is ready

You are helping me ship a piece of writing I've been stuck on. Run the 4-phase methodology: Diagnose, Outline, Draft, Polish. Output the final polished draft only. THE TASK: A blog post announcing our auto-scaling rebuild for our existing customer base. Around 500 words. Lead with the "why," not the "what." We had a couple of high-profile downtime incidents in Q1 that we already wrote post-mortems for... CONTEXT: - Audience: Our customers. Mostly engineering leaders at mid-market SaaS. - Tone: Friendly-direct. Plain-language. Not pitchy. - Length: ~500 words - The hard part: I keep writing it like a press release... ===== MY WRITING STYLE (FOR PATTERN, NEVER FOR COPYING) ===== Sample 1: Most companies I've worked with have one piece of infrastructure that holds them hostage... Sample 2: Spent three hours yesterday in a sprint planning call where someone said... The samples above are STYLE references. Match register, sentence rhythm, opener pattern, contraction usage, hedging tolerance. Do NOT copy exact phrases. ===== HUMANIZE PASS (REQUIRED) ===== After the draft, scan for and remove: - Em-dash clusters (cap: ~1 per 200 words) - Hedging words [full filler-word list embedded] - AI-lecture transitions [full transition-phrase list embedded] - Word-soup adjectives [full list embedded] - Paragraph-announcing openers ("In this post...", "Now let's talk about...") - Stacked abstract noun chains ===== OUTPUT RULES (STRICT) ===== - Output only the polished draft. Nothing else. - No preamble. No clarifying questions. No meta-commentary. - No section labels in the draft (no "Phase 1:", no "Diagnose:") - Only the finished piece, ready to paste into a CMS.
Step 05

Paste into your AI. The polished draft comes back in your writing style.

Open a fresh chat in ChatGPT, Claude, or Gemini. Paste. Hit send. The draft comes back in 30 to 60 seconds, sentence rhythm calibrated to the samples, no AI-tell phrasings, ready to paste into your CMS.

Claude Fresh chat. Polished draft received.

We rebuilt how we scale. Here's why we waited so long.

Most of you saw the post-mortem from January. The one where I had to explain why the dashboard went down for forty minutes during the busiest morning of the quarter. I owe you the followup.

The short version: we were running on a cluster designed for half our current load. Adding capacity worked, but only if someone was watching, and only if they caught it in the first ten minutes. Three times in Q1 nobody caught it in the first ten minutes.

Today we're shipping the rebuild. Auto-scaling is on for every account. It pre-warms based on patterns we've seen in your account, not generic load curves. It scales down too. We don't want to charge you for capacity you're not using.

Two things we explicitly didn't do:

  • We didn't move to a new vendor. The work was about how we use what we have, not who we use.
  • We didn't promise zero-downtime in our SLA. We're at 99.95% measured this quarter. The rebuild puts us in the 99.99% range, but I'd rather you trust the number than trust the marketing.

If you're on a custom plan, I'll be in touch this week to walk through what changes for your specific setup. If you're on the standard tier, you don't need to do anything. The new behavior is on.

The boring answer to "why did this take six months." We couldn't justify the engineering time until we'd lost the same customer twice for the same reason. That happened in February. We started the rebuild the next Monday.

I'd rather have a smaller list of things I can promise than a longer list of things I might.

[Your name]

That's the whole flow.

From a blank doc and a stuck morning to a polished 500-word draft in your writing style, in under 5 minutes of your time.

The 15% that matters (fact-checking, the moments where you sound like you, the personal callouts no AI knows) is yours.

$15
One-time. Lifetime updates. 7-day refund if the product doesn't work as described.
Powered by Gumroad
Walkthrough captured on v1.3.