README.md•14.4 kB
# PitchLense MCP - Professional Startup Risk Analysis Package
[](https://www.python.org/downloads/)
[](https://pitchlense-mcp.readthedocs.io/)
[](https://opensource.org/licenses/MIT)
[](https://pypi.org/project/pitchlense-mcp/)
[](https://github.com/connectaman/Pitchlense-mcp/actions)
A comprehensive Model Context Protocol (MCP) package for analyzing startup investment risks using AI-powered assessment across multiple risk categories. Built with FastMCP and Google Gemini AI.
PitchLense is a comprehensive AI-powered startup analysis platform that provides detailed risk assessment and growth potential evaluation for early-stage ventures. The platform analyzes multiple dimensions of startup risk and provides actionable insights for investors, founders, and stakeholders.
## 🔗 Quick Links
- Website Link : https://www.pitchlense.com
- Web App Github Repo: https://github.com/connectaman/PitchLense
<div align="center">
[](https://youtu.be/XUuLeXaEIdI)
[](https://www.pitchlense.com/)
[](https://github.com/connectaman/PitchLense)
[](https://github.com/connectaman/Pitchlense-mcp)
[](https://pypi.org/project/pitchlense-mcp/)
[](https://pitchlense-mcp.readthedocs.io/en/latest/api.html)
</div>
### 📖 How to Use PitchLense
Watch our comprehensive tutorial video to learn how to use PitchLense effectively:
[](https://youtu.be/XUuLeXaEIdI)
**Click the image above to watch the tutorial on YouTube**
## 🚀 Features
### Individual Risk Analysis Tools
- **Market Risk Analyzer** - TAM, growth rate, competition, differentiation
- **Product Risk Analyzer** - Development stage, market fit, technical feasibility, IP protection
- **Team Risk Analyzer** - Leadership depth, founder stability, skill gaps, credibility
- **Financial Risk Analyzer** - Metrics consistency, burn rate, projections, CAC/LTV
- **Customer Risk Analyzer** - Traction levels, churn rate, retention, customer concentration
- **Operational Risk Analyzer** - Supply chain, GTM strategy, efficiency, execution
- **Competitive Risk Analyzer** - Incumbent strength, entry barriers, defensibility
- **Legal Risk Analyzer** - Regulatory environment, compliance, legal disputes
- **Exit Risk Analyzer** - Exit pathways, sector activity, late-stage appeal
### Comprehensive Analysis Tools & Data Sources
- **Comprehensive Risk Scanner** - Full analysis across all risk categories
- **Quick Risk Assessment** - Fast assessment of critical risk areas
- **Peer Benchmarking** - Compare metrics against sector/stage peers
- **SerpAPI Google News Tool** - Fetches first-page Google News with URLs and thumbnails
- **Perplexity Search Tool** - Answers with cited sources and URLs
## 📊 Risk Categories Covered
| Category | Key risks |
| --- | --- |
| Market | Small/overstated TAM; weak growth; crowded space; limited differentiation; niche dependence |
| Product | Early stage; unclear PMF; technical uncertainty; weak IP; poor scalability |
| Team/Founder | Single-founder risk; churn; skill gaps; credibility; misaligned incentives |
| Financial | Inconsistent metrics; high burn/short runway; optimistic projections; unfavorable CAC/LTV; low margins |
| Customer & Traction | Low traction; high churn; low retention; no marquee customers; concentration risk |
| Operational | Fragile supply chain; unclear GTM; operational inefficiency; poor execution |
| Competitive | Strong incumbents; low entry barriers; weak defensibility; saturation |
| Legal & Regulatory | Grey/untested areas; compliance gaps; disputes; IP risks |
| Exit | Unclear pathways; low sector exit activity; weak late‑stage appeal |
## 🛠️ Installation
### From PyPI (Recommended)
```bash
pip install pitchlense-mcp
```
### From Source
```bash
git clone https://github.com/pitchlense/pitchlense-mcp.git
cd pitchlense-mcp
pip install -e .
```
### Development Installation
```bash
git clone https://github.com/pitchlense/pitchlense-mcp.git
cd pitchlense-mcp
pip install -e ".[dev]"
```
## 🔑 Setup
### 1. Get Gemini API Key
1. Visit [Google AI Studio](https://makersuite.google.com/app/apikey)
2. Create a new API key
3. Copy the API key
### 2. Create .env
```bash
cp .env.template .env
# edit .env and fill in keys
```
Supported variables:
```
GEMINI_API_KEY=
SERPAPI_API_KEY=
PERPLEXITY_API_KEY=
```
## 🚀 Usage
### Command Line Interface
#### Run Comprehensive Analysis
```bash
# Create sample data
pitchlense-mcp sample --output my_startup.json
# Run comprehensive analysis
pitchlense-mcp analyze --input my_startup.json --output results.json
```
#### Run Quick Assessment
```bash
pitchlense-mcp quick --input my_startup.json --output quick_results.json
```
#### Start MCP Server
```bash
pitchlense-mcp server
```
### Python API
#### Basic Usage (single text input)
```python
from pitchlense_mcp import ComprehensiveRiskScanner
# Initialize scanner (reads GEMINI_API_KEY from env if not provided)
scanner = ComprehensiveRiskScanner()
# Provide all startup info as one organized text string
startup_info = """
Name: TechFlow Solutions
Industry: SaaS/Productivity Software
Stage: Series A
Business Model:
AI-powered workflow automation for SMBs; subscription pricing.
Financials:
MRR: $45k; Burn: $35k; Runway: 8 months; LTV/CAC: 13.3
Traction:
250 customers; 1,200 MAU; Churn: 5% monthly; NRR: 110%
Team:
CEO: Sarah Chen; CTO: Michael Rodriguez; Team size: 12
Market & Competition:
TAM: $12B; Competitors: Zapier, Power Automate; Growth: 15% YoY
"""
# Run comprehensive analysis
results = scanner.comprehensive_startup_risk_analysis(startup_info)
print(f"Overall Risk Level: {results['overall_risk_level']}")
print(f"Overall Risk Score: {results['overall_score']}/10")
print(f"Investment Recommendation: {results['investment_recommendation']}")
```
#### Individual Risk Analysis (text input)
```python
from pitchlense_mcp import MarketRiskAnalyzer, GeminiLLM
# Initialize components
llm_client = GeminiLLM(api_key="your_api_key")
market_analyzer = MarketRiskAnalyzer(llm_client)
# Analyze market risks
market_results = market_analyzer.analyze(startup_info)
print(f"Market Risk Level: {market_results['overall_risk_level']}")
```
### MCP Server Integration
The package provides a complete MCP server that can be integrated with MCP-compatible clients:
```python
from pitchlense_mcp import ComprehensiveRiskScanner
# Start MCP server
scanner = ComprehensiveRiskScanner()
scanner.run()
```
## 📋 Input Data Format
The primary input is a single organized text string containing all startup information (details, metrics, traction, news, competitive landscape, etc.). This is the format used by all analyzers and MCP tools.
Example text input:
```
Name: AcmeAI
Industry: Fintech (Lending)
Stage: Seed
Summary:
Building AI-driven credit risk models for SMB lending; initial pilots with 5 lenders.
Financials:
MRR: $12k; Burn: $60k; Runway: 10 months; Gross Margin: 78%
Traction:
200 paying SMBs; 30% MoM growth; Churn: 3% monthly; CAC: $220; LTV: $2,100
Team:
Founders: Jane Doe (ex-Square), John Lee (ex-Stripe); Team size: 9
Market & Competition:
TAM: $25B; Competitors: Blend, Upstart; Advantage: faster underwriting via proprietary data partnerships
```
Tip: See `examples/text_input_example.py` for a complete end-to-end script and JSON export of results.
## 📊 Output Format
All tools return structured JSON responses with:
```json
{
"startup_name": "Startup Name",
"overall_risk_level": "low|medium|high|critical",
"overall_score": 1-10,
"risk_categories": [
{
"category_name": "Risk Category",
"overall_risk_level": "low|medium|high|critical",
"category_score": 1-10,
"indicators": [
{
"indicator": "Specific risk factor",
"risk_level": "low|medium|high|critical",
"score": 1-10,
"description": "Detailed risk description",
"recommendation": "Mitigation action"
}
],
"summary": "Category summary"
}
],
"key_concerns": ["Top 5 concerns"],
"investment_recommendation": "Investment advice",
"confidence_score": 0.0-1.0,
"analysis_metadata": {
"total_categories_analyzed": 9,
"successful_analyses": 9,
"analysis_timestamp": "2024-01-01T00:00:00Z"
}
}
```
## 🎯 Use Cases
- **Investor Due Diligence** - Comprehensive risk assessment for investment decisions
- **Startup Self-Assessment** - Identify and mitigate key risk areas
- **Portfolio Risk Management** - Assess risk across startup portfolio
- **Accelerator/Incubator Screening** - Evaluate startup applications
- **M&A Risk Analysis** - Assess acquisition targets
- **Research & Analysis** - Academic and industry research on startup risks
## 🏗️ Architecture
### Package Structure
```
pitchlense-mcp/
├── pitchlense_mcp/
│ ├── __init__.py
│ ├── cli.py # Command-line interface
│ ├── core/ # Core functionality
│ │ ├── __init__.py
│ │ ├── base.py # Base classes
│ │ ├── gemini_client.py # Gemini AI integration
│ │ └── comprehensive_scanner.py
│ ├── models/ # Data models
│ │ ├── __init__.py
│ │ └── risk_models.py
│ ├── analyzers/ # Individual risk analyzers
│ │ ├── __init__.py
│ │ ├── market_risk.py
│ │ ├── product_risk.py
│ │ ├── team_risk.py
│ │ ├── financial_risk.py
│ │ ├── customer_risk.py
│ │ ├── operational_risk.py
│ │ ├── competitive_risk.py
│ │ ├── legal_risk.py
│ │ └── exit_risk.py
│ └── utils/ # Utility functions
├── tests/ # Test suite
├── docs/ # Documentation
├── examples/ # Example usage
├── setup.py
├── pyproject.toml
├── requirements.txt
└── README.md
```
### Key Components
1. **Base Classes** (`core/base.py`)
- `BaseLLM` - Abstract base for LLM integrations
- `BaseRiskAnalyzer` - Base class for all risk analyzers
- `BaseMCPTool` - Base class for MCP tools
2. **Gemini Integration** (`core/gemini_client.py`)
- `GeminiLLM` - Main LLM client
- `GeminiTextGenerator` - Text generation
- `GeminiImageAnalyzer` - Image analysis
- `GeminiVideoAnalyzer` - Video analysis
- `GeminiAudioAnalyzer` - Audio analysis
- `GeminiDocumentAnalyzer` - Document analysis
3. **Risk Analyzers** (`analyzers/`)
- Individual analyzers for each risk category
- Consistent interface and output format
- Extensible architecture
4. **Models** (`models/risk_models.py`)
- Pydantic models for type safety
- Structured data validation
- Clear data contracts
## 🔧 Development
### Setup Development Environment
```bash
git clone https://github.com/pitchlense/pitchlense-mcp.git
cd pitchlense-mcp
pip install -e ".[dev]"
pre-commit install
```
### Run Tests
```bash
# Create and activate a virtual environment (recommended)
python3 -m venv .venv
source .venv/bin/activate
# Install dev extras (pytest, pytest-cov, linters)
pip install -e ".[dev]"
# Run tests with coverage and avoid global plugin conflicts
PYTEST_DISABLE_PLUGIN_AUTOLOAD=1 pytest -q -p pytest_cov
```
Notes:
- Coverage reports are written to `htmlcov/index.html` and `coverage.xml`.
- If you see errors about unknown `--cov` options, ensure you passed `-p pytest_cov` when `PYTEST_DISABLE_PLUGIN_AUTOLOAD=1` is set.
### Example Scripts
```bash
python examples/basic_usage.py
python examples/text_input_example.py
```
### Code Formatting
```bash
black pitchlense_mcp/
flake8 pitchlense_mcp/
mypy pitchlense_mcp/
```
### Build Package
```bash
python -m build
```
## 📝 Notes
- All risk scores are on a 1-10 scale (1 = lowest risk, 10 = highest risk)
- Risk levels: low (1-3), medium (4-6), high (7-8), critical (9-10)
- Individual tools can be used independently or combined for comprehensive analysis
- The system handles API failures gracefully with fallback responses
- All tables and structured data are returned in JSON format
- Professional package architecture with proper separation of concerns
## 🤝 Contributing
We welcome contributions! Please see our [Contributing Guide](CONTRIBUTING.md) for details.
1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Add tests
5. Submit a pull request
## 📄 License
This project is licensed under the MIT License - see the [LICENSE](LICENSE) file for details.
## 🆘 Support
- **Documentation**: [https://pitchlense-mcp.readthedocs.io/](https://pitchlense-mcp.readthedocs.io/)
- **Issues**: [GitHub Issues](https://github.com/pitchlense/pitchlense-mcp/issues)
- **Email**: connectamanulla@gmail.com
## 🙏 Acknowledgments
- Google Gemini AI for providing the underlying AI capabilities
- FastMCP for the Model Context Protocol implementation
- The open-source community for inspiration and tools
---
**PitchLense MCP** - Making startup risk analysis accessible, comprehensive, and AI-powered.