ganeshkota/ganesh-kota-mcp-revenue-leak-copilot

MCP Revenue Leak Copilot

A robust, business-impactful MCP project that identifies and prioritizes SaaS revenue leakage risks.

Business Value

This project targets high-impact revenue leakage sources:

• —overdue receivables
• —unbilled product usage
• —high-risk renewals

Outputs are structured for action: each risk includes estimated dollars at risk, evidence, and recommended next steps.

Architecture

• —MCP servers:
• —billing_server
• —usage_server
• —contract_server
• —crm_server
• —MCP client: multi-server discovery and routing
• —orchestrator agent: executes scan workflows and generates final report
• —evaluation hooks: checks output quality and actionability

Detailed architecture: see docs/architecture.md.

Quick Start

1. 1.Create and activate a virtual environment.
2. 2.Install dependencies:

pip install -r requirements.txt

1. 1.Run the demo:

python run_demo.py --days 60 --min-at-risk-usd 1000

Example Output

• —ranked risky accounts
• —total at-risk revenue in USD
• —recommended action playbooks per risk type
• —evaluation status

Repository Layout

• —src/revenuecopilot/mcpservers: business system MCP servers
• —src/revenuecopilot/mcpclient: multi-server MCP client
• —src/revenue_copilot/agent: orchestration and evaluation
• —data: realistic sample datasets
• —tests: unit tests and integration-style checks

Testing

pytest -v

Linting

ruff check .

GitHub Ready Checklist

• —clean packaging via pyproject.toml
• —reproducible local install via requirements.txt
• —CI workflow for lint + tests
• —typed Pydantic models for report contracts

License

MIT (add LICENSE file if required by your org policy)