killmyidea
Startup-idea evaluator where Jev scores eight indie-hacker questions and code turns the weighted average into KILL, FIX or SHIP.
Overview
Kill My Idea takes a description of a startup idea and returns a verdict: kill it, fix it or ship it, with no generative LLM. One request asks Jev ten questions in parallel: eight plain indie-hacker questions scored 0 to 4, plus the idea's category and whether it is understandable. Each score is multiplied by 25, a weighted average is computed, a clarity gate applies, and the result maps to KILL below 50, FIX from 50 to 64 and SHIP at 65 or above.
Weights change with the stated goal: making money weights the real problem and money questions twice, open source replaces money with adoption, and just for fun swaps in immediate appeal and fun. A collapsible 'How Jev decided' panel shows raw answers, probabilities, confidence, the average, latency and token use. The key stays server-side, evaluations are archived in SQLite with a per-request opt-out, and a mock mode supports UI work without a key. It includes unit tests and a balanced scoring benchmark script, and deploys to Vercel.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Score | How strongly does this idea solve a real problem / make money / etc.? (eight questions, 0-4) | 0 · 1 · 2 · 3 · 4 |
| Choice | What category is this idea? | — |
| Yes / no | Is the idea understandable? | — |
How it works
- User describes the idea and picks a goal
- Jev answers ten questions in one request
- Scores are scaled, weighted by goal and averaged
- Clarity gate applies, then thresholds give KILL, FIX or SHIP
- Raw answers are shown in a debug panel
Strengths
- Transparent scoring code and thresholds
- Goal-specific weighting
- Mock mode and tests for development
Caveats
- Verdicts are model judgments, not market validation
- Archived evaluations need persistent storage
- Requires a TypeSafe key for live use
Getting started
npm install, cp .env.example .env (set TYPESAFE_API_KEY or TYPESAFE_MOCK=1), npm run dev.At a glance
- Type
- App
- Status
- prototype
- Language
- TypeScript
- Runtime
- Vite app with Vercel or Node API
- Has code
- yes
- Has tests
- yes
Use cases and integrations
How curated lists describe it
A startup-idea evaluation demo assigning KILL, FIX or SHIP labels from Jev scores.
logicrw/awesome-jev-projectsStartup-idea evaluator that chooses kill, fix, or ship.
cobanov/awesome-jevDescribe your startup idea. Jev decides: kill it, fix it or ship it.
hellogumbo/awesome-jev
Frequently asked questions
How is the verdict computed?
Eight 0-4 scores are scaled, weighted by your goal and averaged; below 50 is KILL, 50-64 FIX, 65+ SHIP.
Does it use an LLM?
No, only Jev's typed answers and code.
Can I see why it decided?
Yes, a panel shows raw answers, probabilities and confidence.
Related projects
Secure, persistent VM sandboxes for AI agents, with a PR-review example that compares a general model with Jev on candidate findings.
ProjectproductionPythonChecks whether each cited paper supports its citing sentence: code fetches papers, Claude proves a quote, Jev scores it, a human decides.
ServiceproductionClaude Code plugin that logs your prompts locally and later has Jev score how well you prompt the coding agent over time.
PluginbetaTypeScriptsort where the key is a description: Jev judges pairs of texts and a Bradley-Terry fit turns them into a scored ranking.
ProjectbetaPythonOh My Pi skill for method-first bulk classification: decide when to use Jev, then pilot and freeze a rubric for judge_batch.
Agent skillbetaPythonChrome extension that grades readable page sections on ten rubric scores with Jev via Vercel AI Gateway and gives the page an A-E grade.
ProjectbetaTypeScript