DEV Community

Alex Spinov
Alex Spinov

Posted on

I Replaced 5 SaaS Tools with One MCP Server (Here's How)

#ai #productivity #opensource #tutorial

Last month my SaaS bill was $247/month. Today it's $0.

Not because I stopped using the tools. Because I replaced them with a single MCP (Model Context Protocol) server that Claude talks to directly.

Here's what I replaced and how.

What is MCP?

MCP is an open protocol that lets AI assistants (like Claude) connect to external tools and data sources. Think of it as USB-C for AI — one standard connector for everything.

Instead of switching between 5 browser tabs, I tell Claude: "Check my database, create a report, send it to Slack." One sentence. Done.

What I Replaced

SaaS Tool Cost/mo MCP Replacement Setup Time
Postman Teams $49 MCP API tester 20 min
Datadog (basic) $71 MCP log analyzer 30 min
Notion API $48 MCP + SQLite 15 min
Zapier $49 MCP + webhooks 25 min
Airtable Pro $30 MCP + JSON files 10 min
Total $247 $0 100 min

The Architecture 

Claude (AI) ←→ MCP Server ←→ Your tools/data
                  ↓
         ┌───────┼───────┐
         ↓       ↓       ↓
      SQLite  Webhooks  APIs
      (data)  (actions) (external)
Enter fullscreen mode Exit fullscreen mode

Building the MCP Server

Here's the skeleton:

from mcp.server import Server
from mcp.types import Tool, TextContent
import sqlite3
import httpx

server = Server("my-workspace")
db = sqlite3.connect('workspace.db')

@server.list_tools()
async def list_tools():
    return [
        Tool(
            name="query_data",
            description="Query the workspace database",
            inputSchema={
                "type": "object",
                "properties": {
                    "sql": {"type": "string", "description": "SQL query"}
                },
                "required": ["sql"]
            }
        ),
        Tool(
            name="check_api",
            description="Test an API endpoint",
            inputSchema={
                "type": "object",
                "properties": {
                    "url": {"type": "string"},
                    "method": {"type": "string", "default": "GET"}
                },
                "required": ["url"]
            }
        ),
        Tool(
            name="send_webhook",
            description="Send data to a webhook URL",
            inputSchema={
                "type": "object",
                "properties": {
                    "url": {"type": "string"},
                    "data": {"type": "object"}
                },
                "required": ["url", "data"]
            }
        )
    ]

@server.call_tool()
async def call_tool(name: str, arguments: dict):
    if name == "query_data":
        cursor = db.execute(arguments["sql"])
        rows = cursor.fetchall()
        columns = [d[0] for d in cursor.description] if cursor.description else []
        result = [{col: val for col, val in zip(columns, row)} for row in rows]
        return [TextContent(type="text", text=str(result))]

    elif name == "check_api":
        async with httpx.AsyncClient() as client:
            method = arguments.get("method", "GET")
            r = await client.request(method, arguments["url"], timeout=30)
            return [TextContent(type="text", text=f"Status: {r.status_code}\nTime: {r.elapsed.total_seconds():.2f}s\nBody: {r.text[:500]}")]

    elif name == "send_webhook":
        async with httpx.AsyncClient() as client:
            r = await client.post(arguments["url"], json=arguments["data"])
            return [TextContent(type="text", text=f"Sent. Status: {r.status_code}")]
Enter fullscreen mode Exit fullscreen mode

Replacing Each Tool

Postman → MCP API Tester

Before: Open Postman → find collection → set headers → send → read response.

After: "Claude, test the /users endpoint with this auth token."

Datadog → MCP Log Analyzer 

@server.call_tool()
async def analyze_logs(name, arguments):
    if name == "analyze_logs":
        log_file = Path(arguments["path"])
        lines = log_file.read_text().splitlines()[-1000:]  # Last 1000 lines
        errors = [l for l in lines if 'ERROR' in l]
        warnings = [l for l in lines if 'WARN' in l]
        return [TextContent(type="text", 
            text=f"Last 1000 lines: {len(errors)} errors, {len(warnings)} warnings\n"
                 f"Recent errors:\n" + "\n".join(errors[-5:]))]
Enter fullscreen mode Exit fullscreen mode

Notion → MCP + SQLite

Notion's API is slow. SQLite is instant. I migrated my project data to SQLite and now Claude queries it directly.

Zapier → MCP Webhooks

Instead of visual workflow builders, I just tell Claude: "When the API returns status 500, send an alert to Slack."

Is This For Everyone?

Yes, if:

  • You're a developer comfortable with Python/TypeScript
  • You use 3+ SaaS tools for data/automation
  • You already use Claude or another MCP-compatible AI

No, if:

  • You need enterprise features (SSO, audit logs)
  • You're on a team that needs shared dashboards
  • You're not comfortable running your own server

Resources

Have you tried building an MCP server? What SaaS would you replace first? 👇

Building developer tools at github.com/Spinov001

Top comments (0)