🪦 Idea Autopsy — a Claude Code skill that kills bad business ideas before you build them

Most business ideas deserve to die before you spend a dollar on them. This skill turns Claude Code into a ruthless idea pathologist: kill-list check → five hard filters → the free-AI one-prompt test → live-market verification with your own eyes → a verdict with a named kill-pattern.

Built from a real system that killed 42 business ideas (including a 9/10-scored idea and one that turned out to be federally illegal) before a single one was built. I post the autopsies here: YouTube @actyte · Instagram @hafiz.actyte · LinkedIn.

Install

As a Claude Code plugin (recommended):

/plugin marketplace add hafiz-actyte/idea-autopsy
/plugin install idea-autopsy@idea-autopsy

Or plain clone (works for any harness that reads skills folders):

git clone https://github.com/hafiz-actyte/idea-autopsy /tmp/idea-autopsy \
  && cp -r /tmp/idea-autopsy/skills/idea-autopsy ~/.claude/skills/

Then, in any project:

autopsy my idea: <describe the idea>

What it does

1. Kill-list check — matches the idea against your own REJECTION.md graveyard (starter template included). Dead ideas stay dead; no re-research.
2. Five filters — real pain? buyer has money now? proven demand? legal to charge for? a moat? One hard NO = dead.
3. Free-AI test — if one prompt produces your whole deliverable, you don't have a product. You have a prompt.
4. Own-eyes verification — a Meta Ad Library checklist that catches the traps desk research misses: zero-ads categories, incumbent-owned lanes, and the "1,100 advertisers means demand" trap (it also means everyone could build it — so everyone did).
5. Verdict — DEAD with a named kill-pattern (14 documented in references/kill-patterns.md), or SURVIVED with the one cheapest test that could still kill it. Every autopsy writes a row to your kill-list.

Why kill ideas?

Building was never the problem — AI builds anything in a weekend now. Picking is the problem. A fast, honest kill is a win: it costs 20 minutes instead of 3 months. The kill-list compounds — every dead idea makes the next autopsy faster.

Files

• skills/idea-autopsy/SKILL.md — the autopsy procedure Claude follows
• skills/idea-autopsy/references/kill-patterns.md — the 14 named ways ideas die
• skills/idea-autopsy/templates/REJECTION-template.md — your starter kill-list

Watch ideas die

Real autopsies from the 42-idea graveyard, one per week — comment your idea; the most-liked one gets autopsied on camera:

• YouTube: @actyte
• Instagram: @hafiz.actyte
• LinkedIn: Hafiz Siddiq

MIT license. PRs welcome — especially new kill-patterns with receipts.