Once you're comfortable asking the MCP questions, the natural next step is asking it to make something. Three things come up most often: reports you can hand to someone, dashboards you can look at, and workflows that pull Sitebulb together with the other tools you already use.
None of this needs a template or a setup step. You describe what you want, read what comes back, and refine. Treat the first answer as a draft you shape, not a finished article.
Not connected yet? Everything on this page assumes the Sitebulb MCP is already hooked up to Claude or ChatGPT. If it isn't, Sitebulb MCP: Start Here gets you connected in a couple of minutes, then Getting Started with the Sitebulb MCP covers your first prompts.
Reports
The audit already holds the findings. A report is just those findings written up for a particular reader, and that's a job the MCP is good at: turning crawl data into something a client, a manager or a developer can actually read.
You can ask for very different reports off the same crawl:
A plain site-health summary: "Summarise the latest [project] audit in plain English."
A client-ready write-up with an executive summary, the priorities and the quick wins.
A developer-facing list: the issues, why they matter, and example URLs to check.
Ask for the format you need, too. By default the report lands in the chat, but you can ask for it as Markdown to paste elsewhere, or as a Word document or Google Doc (if you've connected it to your assistant).
The more comprehensive you are about what you want, the more accurate the output will be. Here's a worked example:
"Using the Sitebulb MCP, take the latest audit for [Project name] and tell me the biggest issues and which I should fix first. Give me the answer as a prioritised summary table only, with columns for Priority, Cluster (group related hints into fix clusters), Scale (how much of the site it affects), Effort, and Owner."
Note: The MCP works from a sample of up to 50 example URLs per Hint, so a report gives you the right shape and the headline numbers, not an exhaustive URL-by-URL list. When you need every affected URL, pull the export from the Sitebulb app.
Dashboards
Sometimes you don't want prose, you want to see it. The crawl data comes back as clean, structured numbers, so an assistant can turn it straight into charts and simple dashboards without you touching a spreadsheet.
A few things work particularly well:
A single-audit overview: the site score, the split of issues by severity, and the breakdown by category as a bar or pie chart.
A trend over time: site score and key counts across your last several crawls, so you see the direction of travel rather than one snapshot. This one draws on a project's crawl history, so the more crawls it has, the richer the picture.
A wins-and-regressions board comparing the two most recent crawls, colour-coded, for a client update.
Where the dashboard appears depends on your assistant, but you can usually view it inline and save or screenshot it for a report or a client folder. Here's a worked example:
"Using the Sitebulb MCP, pull the last 12 weekly crawls for [Project name] and build me an HTML dashboard showing the audit trend over time. Across the top, show summary cards for the latest site score, Critical count, High count, and broken (404) count - each with the change versus 12 weeks ago. Below that, plot the site score across all 12 crawls as a filled line chart. Make it look polished, like something I could put in front of a client."
Combining Sitebulb with your other tools
This is where it gets interesting, and it's worth being clear about how it works. The Sitebulb MCP only ever pulls back Sitebulb data. Your AI assistant can combine this with other tools because it can also talk to the other tools you've connected to it, and reason across all of them in one conversation.
So if you've connected something like Ahrefs, Google Search Console, Jira or Slack to the same assistant, you can ask things no single tool could answer on its own:
Add commercial weight to the findings: "Cross-reference the broken pages in the latest [project] audit with their traffic from [GSC], and prioritise by what's costing us the most visits."
Turn issues into action: "Turn the top five issues from the latest [project] audit into Jira tickets."
Get it in front of people: "Summarise the latest [project] audit and post the headlines to our Slack channel."
Once you start connecting the other tools in your stack, you'll start to see how the MCP can power an entire ecosystem.
For example, I have this Claude Cowork prompt running on a weekly Sitebulb crawl:
I have a crawl set to run on a weekly schedule already in Sitebulb. Claude automates the MCP pull from Sitebulb to grab the latest data, it runs analysis and then pings me in Slack to let me know if there are any issues I need to watch out for:
This is essentially the 'website monitoring' feature we get asked to build all the time.
As another example, we use Jira internally as our PM tool, and so when I find website issues I need to pass across to dev, I can pull the hint details out of Sitebulb MCP and use it to build Jira tickets - all without leaving Claude:
The limit here is simple: the Sitebulb MCP can't reach those other tools for you. They have to be connected to your assistant, and the quality of the answer depends on the data each one brings.
Next steps
The best way to get a feel for this is to try it - it's better than any amount of reading on the subject. Pick one thing from this page, a report, a dashboard, or a workflow, and run it against a real crawl. Treat the first answer as a draft and shape it until it's useful - and don't forget you can always ask your AI assistant to help figure out what you can do.
When you're ready for the next step up, Skills are where this goes further. A good prompt gets you a good answer; a Skill turns a repeatable job into a fixed recipe, so you get the same shape of output every time without writing the prompt out again. Have a read of the Sitebulb Skills for Claude and ChatGPT article when you want to build that consistency in.