🚀 Startup Idea Validator

An AI-powered startup validation agent built with LangGraph and Streamlit. Describe your startup idea and get instant, data-driven feedback on market landscape, community sentiment, and technical feasibility.

Active link: https://phoenix1454-startup-validation-srcapp-wjnche.streamlit.app

---

Features

• Market Landscape Research — Searches the web via Firecrawl for market size, competitors, and industry analysis.
• Community Sentiment Analysis — Queries Hacker News for relevant discussions, upvotes, and comment activity.
• Technical Feasibility Assessment — Searches GitHub for related open-source projects, star counts, and languages used.
• Conversational Memory — Remembers your idea across follow-up questions in the same session.
• Streaming UI — Real-time thinking indicators and token-by-token response streaming in a Streamlit chat interface.

Architecture

User ── Streamlit Chat UI ── LangGraph ReAct Agent (o4-mini)
                                  │
                    ┌─────────────┼─────────────┐
                    ▼             ▼             ▼
             Firecrawl      Hacker News      GitHub
           (market data)   (sentiment)    (feasibility)

File	Purpose
src/app.py	Streamlit chat interface with streaming
src/validator.py	LangGraph agent setup and system prompt
src/tools.py	Tool functions (Firecrawl, HN, GitHub)

Getting Started

Prerequisites

• Python 3.13+
• uv (recommended) or pip

API Keys

Create a .env file in the project root:

OPENAI_API_KEY=sk-...
FIRECRAWL_API_KEY=fc-...
GITHUB_TOKEN=ghp_...        # optional, increases GitHub rate limit

Installation

# Clone the repo
git clone https://github.com/Phoenix1454/Startup-Validation.git
cd Startup-Validation

# Install dependencies (using uv)
uv sync

# Or with pip
pip install -r pyproject.toml

Run

python -m streamlit run src/app.py

The app will open at http://localhost:8501.

Usage

1. Describe your idea — e.g. "I want to build an AI-powered resume screening tool for small businesses"
2. The agent will immediately research the market landscape and present findings.
3. Ask follow-ups — "What about competitors?", "Check community sentiment", "Assess technical feasibility"
4. Full validation — type "Do a full validation" to run all three tools and get a structured VALIDATE / NEEDS_WORK / REJECT recommendation.

Testing

Unit tests cover the tool functions (Firecrawl, Hacker News, GitHub) with mocked HTTP calls — no API keys required:

uv run pytest tests/ -v

Tests run automatically on every push via GitHub Actions.

Tech Stack

Technology	Role
LangGraph	ReAct agent orchestration with memory
OpenAI o4-mini	LLM reasoning
Firecrawl	Web search & scraping for market data
Streamlit	Chat UI with streaming support
Hacker News API	Community sentiment via Algolia search
GitHub API	Open-source project discovery

License

MIT