VentureScout

Deal Flow CRM & Analyzer for Venture Capital workflows.

Overview

VentureScout is a full-stack tool that automates the early-stage deal screening workflow for Venture Capital. Paste a single startup URL and the system scrapes the site, generates an AI-driven investment memo (high-concept pitch, industry tags, bull/bear case), and drops the deal onto a drag-and-drop Kanban board.

Features

• One-click analysis — submit a URL, get a full AI-generated deal card
• Web scraping — extracts title, meta description, and page content via BeautifulSoup
• AI investment memos — OpenAI generates a high-concept pitch, industry tags, and bull/bear cases
• Kanban board — drag-and-drop deal cards across columns: New → Outreach → Due Diligence → Pass → Invest
• Deal detail modal — click any card to view the full AI analysis
• Auto logos — fetches company logos via Clearbit with Google Favicon fallback
• Persistent storage — all deals saved to SQLite

Security

• API key authentication — all endpoints gated by X-API-Key header (configurable via VS_API_KEY)
• Rate limiting — 60 req/min globally, 10 req/min on /api/analyze (via flask-limiter)
• SSRF prevention — URL validation blocks internal/private IPs, non-HTTP schemes, and unresolvable hostnames
• CORS restriction — configurable allowed origins (defaults to http://localhost:5173)
• Status validation — PATCH /api/deals/:id only accepts valid pipeline statuses
• Prompt-injection mitigation — scraped text is sanitized before being sent to OpenAI
• Response size limits — scraper enforces a 2 MB download cap
• Generic error responses — internal details are logged server-side, never leaked to clients
• Safe defaults — debug mode off, binds to 127.0.0.1

Tech Stack

• Frontend: React (Vite) + Tailwind CSS v4
• Backend: Python 3.11 (Flask)
• Database: SQLite
• AI/LLM: OpenAI API (GPT-4o-mini) with Pydantic structured output
• Scraping: BeautifulSoup4 + Requests

Project Structure

src/
├── venture_scout/          # Flask backend
│   ├── app.py              # API routes & server entry point
│   ├── scraper.py          # Website scraping logic
│   ├── ai_analyzer.py      # OpenAI integration & Pydantic model
│   ├── database.py         # SQLite schema & queries
│   └── security.py         # Auth, URL validation, rate limiting, sanitization
└── frontEnd/               # React frontend
    └── src/
        ├── App.jsx          # Main app (Kanban board, modal, fetch logic)
        ├── main.jsx         # React entry point
        └── index.css        # Tailwind imports

Prerequisites

• Python 3.11+
• Node.js 18+
• uv (Python package manager)
• An OpenAI API key

Installation

# Clone the repository
git clone https://github.com/abhinav-rb/VentureScout.git
cd VentureScout

# Backend setup
uv venv .venv
source .venv/bin/activate
uv pip install -e .

# Initialize the database
python src/venture_scout/database.py

# Frontend setup
npm install --prefix src/frontEnd

Running the App

You need two terminal tabs running simultaneously:

Tab 1 — Start the Flask backend (port 5001):

export OPENAI_API_KEY='sk-...'
source .venv/bin/activate
python src/venture_scout/app.py

Optional environment variables:

Variable	Purpose	Default
VS_API_KEY	API authentication key (unset = auth disabled)	—
VS_CORS_ORIGINS	Comma-separated allowed CORS origins	http://localhost:5173
FLASK_DEBUG	Set to 1 to enable debug mode	0
VS_HOST	Server bind address	127.0.0.1
VS_PORT	Server port	5001

Tab 2 — Start the Vite dev server (port 5173):

npm run dev --prefix src/frontEnd

Then open http://localhost:5173 in your browser.

The Vite dev server proxies all /api requests to the Flask backend automatically.

API Endpoints

All endpoints require an X-API-Key header when VS_API_KEY is set.

• POST /api/analyze — submit a startup URL for AI analysis (rate-limited: 10/min)
• GET /api/deals — list all tracked deals
• PATCH /api/deals/:id — update a deal's status (valid: New, Outreach, Due Diligence, Pass, Invest)

License

MIT