Kill bad business ideas before you waste time building them

Idea Autopsy examines a proposed business, product, or side project before substantial money or building time is committed. It looks for a decisive weakness rather than default encouragement. The outcome is DEAD, SURVIVED, or INCONCLUSIVE according to the available evidence, with a concise explanation and the cheapest test of the assumption that matters most.

The first step checks an existing REJECTION.md personal kill-list. A matching niche provides history, not an automatic verdict. Compare the cited row with current customers, timing, and assumptions. A familiar failure pattern is a strong prior that still needs the relevant check before it can decide the new idea.

The five filters ask whether the problem is real pain, the buyer has money now, there is observable demand, charging is legal in the target market, and the idea has a defensible advantage. Each requires evidence rather than enthusiasm. A substantiated hard failure can kill the idea; unclear buyer, market, or deliverable details require clarification.

Use when asking whether to build something, validating a new venture, requesting an idea autopsy, or before starting market research and build planning. The method is designed to challenge assumptions before implementation excitement displaces buyer questions.

Do not treat the ability to generate a deliverable with a free model as proof that nobody will pay. Test whether quality control, integration, trust, or distribution still create value. Likewise, advertising demand does not establish room for another entrant, and a long-running ad does not prove profit, spend, or attribution.

Input is the idea, intended buyer and market, core deliverable, relevant historical rejections, and evidence gathered through the filters. The output follows this compact verdict format:

VERDICT: DEAD | SURVIVED | INCONCLUSIVE
KILL-PATTERN: <name>            (if dead)
THE ONE SENTENCE: <the single finding that decided it>
EVIDENCE: <2-4 hard facts with sources/numbers>
NEXT: <if survived: the ONE cheapest test that could still kill it>

For live-market checking, the user examines Meta Ad Library or an equivalent in their own browser. The checklist counts active advertisers and notes the oldest running ad. No ads can signal the wrong channel; a few dominant incumbents and a crowded commodity field require different explanations. Other categories may acquire customers through search or app stores.

REJECTION.md is the only project file the workflow modifies. If it exists, results can be appended as dated dead-idea rows or survivors with a pending test. If it does not exist, creation requires consent. When permission is declined, print the proposed row for the user instead of writing a file.

The skill itself performs no shell commands, network calls, or credential handling. Its ad-library work is a user checklist. It supports an evidence-focused decision process and does not replace legal, financial, or professional market research.