# AI Engineering & Operator Quickstart Pack ⚡

[![Python 3.10+](https://img.shields.io/badge/python-3.10+-blue.svg)](https://www.python.org/downloads/)
[![Model Context Protocol](https://img.shields.io/badge/MCP-Compatible-green.svg)](https://modelcontextprotocol.io/)
[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
[![Digital Viking](https://img.shields.io/badge/Store-digitalviking.io-indigo.svg)](https://digitalviking.io)

> Production-grade FastMCP servers, deterministic prompt evaluation guardrails, hybrid vector search checklists, and automation ROI scorecards for AI engineers and technical operators.

---

## 📦 What's Inside the Quickstart Pack

This repository provides open-source architectural blueprints and starter files designed to eliminate the initial 40+ hours of boilerplate when standing up enterprise AI workflows:

1. **`mcp_quickstart_server.py`**: A standalone FastMCP server implementation featuring:
   - In-memory token-bucket rate limiting (prevents runaway token-burning loops).
   - Strict Pydantic v2 input boundary validation.
   - Standardized JSON-RPC 2.0 error payloads and protocol compliance.
2. **`prompt_engineering_guardrails.md`**: The 10 production prompt patterns we enforce across agent deployments (XML boundary encapsulation, delimiter hygiene, negative constraint enforcement).
3. **`ai_workflow_audit_micro_scorecard.xlsx`**: Dynamic financial ROI model evaluating 10 automation workflows with sensitivity analysis.
4. **`rag_production_checklist.md`**: 12-point pre-launch verification checklist for hybrid dense/sparse search, chunking strategies, and cross-encoder reranking.

---

## 🚀 Quickstart

### 1. Clone & Setup Environment

```bash
git clone https://github.com/jonmarrs/mcp-agent-starter-kit.git
cd mcp-agent-starter-kit

# Create virtual environment using uv or standard venv
uv venv
source .venv/bin/activate

# Install dependencies
uv pip install "mcp>=1.0.0" "pydantic>=2.0.0"
```

### 2. Run the FastMCP Server

```bash
python mcp_quickstart_server.py
```

The server starts listening on stdio, ready to connect to Claude Desktop, Cursor, OpenClaw, or custom agent runners.

```json
{
  "mcpServers": {
    "quickstart": {
      "command": "python",
      "args": ["/path/to/mcp_quickstart_server.py"]
    }
  }
}
```

---

## 🛠️ The Architecture Ladder: Choose Your Depth

| Deliverable | Free Quickstart | MCP Tool Registry ($34.99) | Production RAG Kit ($39.99) | Complete Master Suite ($99.00) |
| :--- | :---: | :---: | :---: | :---: |
| **FastMCP Quickstart Server** | ✅ | ✅ | ✅ | ✅ |
| **Prompt Guardrails & ROI Scorecard** | ✅ | ✅ | ✅ | ✅ |
| **Production Tool Registry (12 Tools)** | ❌ | ✅ | ❌ | ✅ |
| **Postgres & SQLite Hybrid Storage** | ❌ | ✅ | ❌ | ✅ |
| **Dense + BM25 Reciprocal Rank Fusion** | ❌ | ❌ | ✅ | ✅ |
| **Cross-Encoder Reranking Engine** | ❌ | ❌ | ✅ | ✅ |
| **Multi-Agent Async Boilerplate** | ❌ | ❌ | ❌ | ✅ |
| **Agency Legal Contracts & Proposal Pack** | ❌ | ❌ | ❌ | ✅ |
| **All 9 Commercial Production Toolkits** | ❌ | ❌ | ❌ | ✅ |

---

## 🎁 Community Welcome Discount

As a member of the developer community, get **20% OFF** any production toolkit or the all-inclusive Master Suite using our launch voucher:

### Voucher Code: **`WELCOME20`**

- **[Browse All Toolkits on digitalviking.io](https://digitalviking.io)**
- **[Download Free AI Quickstart Pack ($0.00 Checkout)](https://digitalviking.io/products/free-ai-engineering-quickstart-pack)**
- **[Unlock the AI Developer Master Suite ($79.20 with code)](https://digitalviking.io/products/the-complete-ai-builder-agency-master-suite)**

---

## 📄 License & Attribution

- **Quickstart Code & Blueprints:** Licensed under the [MIT License](LICENSE).
- **Commercial Toolkits & Contracts:** Sold under Digital Viking Commercial Team Licenses.