Digital Agency Gas Mark 8 Saves 5-7 Hours Weekly Managing 400 Sites With Cloudways MCP

About Gas Mark 8

Gas Mark 8 is a digital agency serving universities, nonprofits, and healthcare educators across the US and UK. The team builds and hosts websites and eLearning platforms for mission-focused organizations, including the NHS, managing 400 websites across 43 servers in three countries.

This year, the agency won a national HSJ Digital Award in England in the Enhancing Workforce Engagement and Productivity through Digital category, recognizing its work with the Greater Manchester Cancer Alliance, a project hosted entirely on Cloudways infrastructure alongside the rest of its client base.

Cloudways serves as Gas Mark 8’s primary hosting partner, a relationship the agency credits to consistent service as its footprint has grown. Cloudways MCP has since changed how the small team monitors and troubleshoots that footprint. Here’s how.

– Gas Mark 8 serves nonprofits, higher education, small businesses, and NHS education, the mission-focused organizations the agency builds and hosts websites for. (Source: gasmark8.com)

The Challenge: Manual Downtime Fixes

Before Cloudways MCP, Gas Mark 8 managed its servers primarily through the platform, handling much of the day-to-day work by hand. Routine tasks like updating software or standing up new sites were manageable one at a time.

The friction multiplied once the agency’s footprint grew to hundreds of sites. Checking CPU levels across more than 40 servers, or pushing updates across all 400, is where the real time loss showed up.

The cost was clearest during incident response. As Mike Richwalsky described it, “We’d see a site was down, log into Cloudways, find the site, go to the server it lives on, then dig into the monitoring or logs area to find the actual issue.”

“Six incidents in a day is 30 minutes spent just figuring out the status of a server or app before we’ve fixed anything.” — Mike Richwalsky

That was time spent purely on diagnosis, before any fix began, often while a client’s site was already down.

The Solution: MCP-Driven Server Monitoring

Gas Mark 8’s first experiment with Cloudways MCP was deliberately low-risk. The team wrote a skill to pull a list of every server and every app running on it, then built a simple profitability checker comparing what a server costs against what a client pays.

– The Cloudways MCP capabilities including server control, security scanning, and monitoring and alerts. (Source: cloudways.com)

Read-only and non-destructive, the test gave the team confidence to move further. Getting there was a matter of trial and error. As Richwalsky put it, “You can iterate fast across a few ideas and land on the one that works best for you.” When something didn’t work, the team noted why and applied the lesson to the next agent.

That iterative process eventually produced internal tools, built using Claude Code, that let anyone on staff find a site, pull server specs, or check for issues in seconds.

The bigger payoff came once MCP was applied to the agency’s daily flood of monitoring data. Cloudways alert emails, WordFence alerts across 400 sites, and a custom Uptime Kuma setup all generate signals constantly.

“We let the AI be the first wave of defense and lean on the Cloudways MCP to pull the critical info we need instantly.” — Mike Richwalsky

Access is tiered deliberately, with agents operating at different trust levels that expand only as they prove reliable. For now, that trust stops short of letting agents create or modify infrastructure, though Richwalsky expects that to change. He described plans to eventually have an agent spin up a new app, follow the team’s setup guidelines, and install its base plugin suite unattended.

The Result: 5-7 Hours Saved Weekly

With MCP handling the first pass of diagnosis, Gas Mark 8 no longer treats its alert volume as a full-time drain.

5-7 hours saved per week that previously went into manually parsing alerts from Cloudways, WordFence, and Uptime Kuma.

Critical alerts now arrive by email and Slack pre-flagged with the affected server’s status.

Read-only agents handle diagnostics instantly, while spinning up or changing infrastructure still goes through a person by hand.

“The tool is only as good as the guardrails you put around it.” — Mike Richwalsky

– A sample Cloudways MCP query returns real-time CPU and memory usage for a server, the kind of instant diagnostic check MCP enables. (Source: cloudways.com)

The team’s rule remains simple: read freely, change carefully, extend trust only where it’s been earned. For a small agency managing hundreds of sites, that balance is what lets them respond to issues like a much larger team would.

Richwalsky’s advice to other agencies considering MCP is practical rather than theoretical: set up a read-only token and start testing a few ideas. That single step, low-risk by design, was what got Gas Mark 8 started too.

Stop Manual Troubleshooting, Diagnose Instantly.

Cloudways MCP saves hours every week by automating the server checks your team used to do by hand.

Try Cloudways MCP