Stop guessing. Start validating.
AI agents that act as your personal venture analyst — from startup idea brainstorming to full validation and go-to-market strategy. Built for indie developers who'd rather validate in 10 minutes than regret in six months. Powered by Claude Code, OpenAI Codex, and Cursor.
⭐ Star this repo if you've ever asked ChatGPT to generate a million dollar idea.
Clone the repo. Open it in your AI tool. Start talking. No setup, no API keys, no commands.
- Indie app developers who want to validate app ideas before building
- Solo founders exploring B2C startup ideas and looking for a data-driven brainstorming partner
- Side-project builders deciding where to invest their limited time
- Aspiring developers with no idea yet who want to discover one worth building
- Anyone tired of building things nobody wants
No installation. No dependencies. No terminal commands. Just clone and go.
- Clone this repository
git clone https://github.com/MaxKmet/idea-validation-agents.git- Open Claude Code and open the project folder
- Start a new chat and say anything — e.g.,
"I want to build a habit tracker for climbers. Is it worth it?" - The agent activates automatically from
CLAUDE.mdand starts the right workflow
- Clone this repository
git clone https://github.com/MaxKmet/idea-validation-agents.git- Install Codex CLI if you haven't yet
npm install -g @openai/codex- Navigate to the project folder and run Codex
cd idea-validation-agents
codex- Say what you want — e.g.,
"Help me find an app idea". The agent readsAGENTS.mdand routes you automatically.
- Clone this repository
git clone https://github.com/MaxKmet/idea-validation-agents.git- Open the folder in Cursor
- Open the AI chat (Cmd+L / Ctrl+L) and describe your situation
Example:"I'm a solo developer with 3 years experience. Help me validate an idea for climbers — a habit tracker with daily streaks and tips." - The agent reads from
.cursor/rules/and activates automatically
That's it. The agent detects your intent and routes you to the right workflow. All analysis results are saved to the
memory/folder so nothing is lost between sessions.
The agent interviews you about your background, skills, and interests — then researches what's actually trending — and generates 7–10 scored app ideas matched specifically to you.
I don't have an app idea yet. Help me find one.
What should I build? I'm a fitness coach with 8k Instagram followers.
I want to find an app idea in the productivity space.
Steps: background interview → builder profile → trend analysis → idea generation → scoring
Methodology: Trend signals are pulled from TikTok Creative Center (hashtag velocity), Reddit (community pain language), App Store (new entrants + review patterns), and Google Trends (search demand). Ideas are filtered against your domain expertise, skills, and distribution advantages from the interview — so you get ideas you can actually build and sell.
Output: Ranked + scored list of app ideas saved to memory/ideas/
Don't want to answer questions? Say "browse topics" to pick from 20 product domains, or "skip" to jump straight to ideas.
Full 9-step validation. Every dimension scored, weighted, and combined into a final verdict — with the single riskiest assumption identified and a concrete experiment to test it before writing any code.
Validate my idea: an AI tool that rewrites your emails to sound more professional.
I want to build a habit tracker for intermittent fasting. Worth it?
Score this — a subscription app that sends meal plans based on your grocery budget.
Steps: trend analysis → competitor mapping → desire scoring → pricing model → distribution analysis → retention prediction → CAC modeling → final score (0–100) → decision memo
Methodology:
- Scoring uses a multiplicative-floor algorithm — one catastrophic weakness kills the overall score, just like in a real startup
- Pricing is estimated via Van Westendorp price sensitivity analysis + desire-premium multipliers (e.g. survival/status desires command 1.3–2× price premium)
- Distribution models viral coefficient (k-factor) across 6 loop types, ASO opportunity via a 5-factor rubric, and creator economy fit
- Competitors are analyzed via systematic App Store search + 1-star/3-star review mining to surface the exact gaps incumbents leave open
- Riskiest Assumption Test (RAT) designs a ≤2-week, ≤$100 behavioral experiment to validate the single assumption most likely to kill the idea
- Pre-mortem (Klein, 2007) imagines the idea failing in 12 months and traces the most probable causes back to scored weaknesses
Output: decision_memo.md — verdict (pursue / test / pivot / drop), strengths, risks, RAT experiment, kill criteria, and your next step
Research a category before committing to any idea. Understand who already owns it, what users hate, and whether the timing is right.
Tell me about the journaling app market.
What's happening in the AI language learning space?
Is the meditation app market still worth entering?
Steps: multi-platform trend analysis → competitor landscape → market size (TAM/SAM/SOM) → distribution channel assessment
Methodology:
- Trend velocity scored across platforms: rising-fast / rising / stable / declining
- Market saturation rated via a 5-factor rubric (competitor count, incumbent dominance, funding activity, keyword saturation, content saturation)
- Market size uses triangulated bottom-up estimation: search volume × intent conversion rate, community size × platform multiplier, and competitor revenue proxies — cross-checked for consistency
- SOM estimates use indie-realistic capture rate benchmarks by app category (e.g. niche productivity: 0.5–2.0% year 1)
Output: Trend intelligence + competitor map + TAM/SAM/SOM estimates saved to memory/market_insights/
Finds the best version of a failing idea instead of abandoning it entirely. Each pivot option changes exactly 1–2 variables — audience, niche, pricing model, or feature emphasis — with a projected score improvement before you commit.
My idea scored 34/100. Should I pivot?
The validation said to pivot. What are my best options?
This isn't working — what should I change about my fitness app idea?
Steps: re-read scores → weakness root cause analysis → 2–3 pivot options with projected scores → re-score best option
Methodology:
- Weaknesses are classified by root cause: structural (can't fix), situational (fixable with time/budget), knowledge-gap (needs more research), or addressable (clear fix exists) — only the latter two generate pivot options
- Each pivot must pass the Same Idea Test: changes 1–2 variables, preserves at least one strong dimension, and has evidence from market_insights or competitor review mining
- Scoring simulation projects how each dimension shifts before full re-scoring
- Effort is estimated and adjusted for founder tier — what's "medium" for a builder is "high" for a beginner
Output: pivot_options.json with ranked pivots, effort estimates, and projected score ranges
All outputs persist in memory/ between sessions.
memory/
├── user_profile.md ← your builder profile (reused across sessions)
├── market_insights/
│ └── fitness-tiktok-2026-04.md ← trend data per niche + platform
└── ideas/
└── habit-tracker-climbers/
├── competitors.json
├── pricing.json
├── scores.json
└── decision_memo.md ← the final verdict
⭐ Star this repo so you don't lose it.
Validate in 10 minutes. Build with confidence.