Claude Business Idea Evaluator: systematic scoring for startup ideas

If you're anything like me, you have a running list of business ideas that grows faster than you can evaluate them. I wanted a systematic way to score and rank ideas rather than relying on gut feeling, so I built the Claude Business Idea Evaluator — a Claude Code workspace that uses what I call the ICEC methodology to objectively assess business concepts.

The ICEC framework

ICEC extends the traditional ICE prioritization model (Impact, Confidence, Ease) by adding a fourth dimension: Contextual Suitability. This is the dimension most evaluation frameworks miss — an idea might be brilliant in the abstract but a terrible fit for your specific life circumstances. Each dimension is scored 1-10, and the final score is the average across all four.

The contextual scoring is what makes this more than a generic framework. You fill out a detailed user context file describing your skills, financial situation, time availability, location, and goals. Claude uses this to ground its assessment in your actual reality rather than evaluating ideas in a vacuum.

How it works in practice

You write up each business idea in a markdown file covering the problem, proposed solution, target market, and any specifics you have. Drop it in the to-process/ folder and run /evaluate-ideas. The specialized evaluator agent reads your idea, conducts web research for market dynamics and competitive landscape, scores each ICEC dimension with detailed reasoning, and generates a structured JSON analysis. All evaluated ideas feed into a central rankings database so you can compare across your portfolio.

There are additional slash commands for viewing rankings, comparing two ideas side-by-side, generating templates, and exporting comprehensive reports. The evaluator agent is configured to be honest — it uses substantive reasoning rather than generic platitudes and will honestly flag weaknesses.

Score interpretation

Scores above 8.0 indicate strong candidates with high viability and alignment. The 6.5-7.9 range means worth exploring but needs validation. Below 5.0 generally means poor fit or weak fundamentals. I've found the framework most useful not for the absolute scores, but for the reasoning behind each dimension — it surfaces considerations I hadn't thought of.

Check out the repo on GitHub. It's open source and designed to be adapted — you can adjust the scoring formula, modify thresholds, or add new evaluation dimensions.