Rank Prompt AI Visibility: Scan, MCP and Fixes
Quick answer: In Rank Prompt, enter a site, run a visibility scan, pick categories so the product can generate prompts, then read where those prompts show up across the models you selected. The 12 November 2026 recording then installed the Rank Prompt MCP in Claude Code, authenticated with an API key, and used that data to edit pages and draft outreach. Rank Prompt sponsored this video. The on-site edits were not treated as an instant rank change.
Rank Prompt sponsored this video. This guide uses a plain product link with no tracking parameters: https://www.rankprompt.com. Credits, plan prices and visibility scores below are what the recording showed on 12 November 2026, not a live quote. Brands considering a similar build can read the sponsorship packages and editorial approach.
This is the recorded Rank Prompt walkthrough, not a generic AI-visibility tour. For a free domain scan and report on Mike's own product, see VisiScan AI Visibility: Free Scan and Report. For Semrush data over MCP, see Semrush MCP: AI Visibility and Keyword Gaps. For crawler hits versus real visits, see Trakkr and Codex: AI Visibility, Crawlers and Visits.
What this recording shows
The opening question is what happens when someone asks ChatGPT to recommend a business like yours, and whether that answer names you or a competitor. Mike's point on camera is that a site can still be missing from AI recommendations even if it ranks on Google. Rank Prompt is the tool used to check that gap.
He names ChatGPT, Gemini, Claude, Perplexity, Grok and Muse as the models the platform can show visibility across. The walkthrough analyzes VisiScan, finds prompts where it is missing, then uses Claude Code with the Rank Prompt MCP to edit content and send outreach. If MCP itself is new, start with What is MCP? The Model Context Protocol explained. For Claude Code install, see Claude Code tutorial for beginners: install and use it.
1. Run a free AI visibility scan
On the Rank Prompt site he enters VisiScan, sets the country to the United States and the language to English, skips the client-contact fields, and creates the brand. The dashboard greets the Automajic account, says VisiScan has no report yet, and recommends running the first one. He clicks run the first report.
ChatGPT, Claude and AI mode are on by default in this recording. He also turns on Grok, Perplexity, Gemini and Muse, then clicks check visibility. The UI validates the URL and creates the report. On screen it analyzes online presence, identifies brand categories, detects brand aliases and prepares AI insights. He says that pass should take about 30 to 40 seconds. Treat that timing as what he said on 12 November 2026, not a guarantee.
2. Track prompts and LLM categories
After the first report he picks categories to track: AI citation tracking, answer engine analysis, generative engine optimization tools, AI search monitoring, AI visibility software and LLM brand tracking. He says Rank Prompt then generates 15 prompts for each category. That generation, on camera, took about 5 to 10 seconds.
One decision prompt shown in the list is:
which AI visibility measurement tools offer a free scan before I pay for reportThe list also groups consideration and awareness prompts. Starting the visibility analysis in this session cost 45 credits. He said he had about 1,000 credits per month on the account, and that an upgrade on screen went up to about $299 per month. Those numbers are from 12 November 2026. Check Rank Prompt for current credits and plans.
3. Read visibility and competitor rank
The report then walks the selected prompts across ChatGPT, Claude, Perplexity, Google AI mode, Grok and Muse. He can filter to one category, such as AI citation tracking, or turn all of them on. For VisiScan in this recording, overall visibility looked poor. On the competitors tab the site sat at 155 out of 157, with share of voice at 0.1%.
Most tracked prompts showed about 0% visibility. One prompt scored 14% overall, and he said VisiScan was showing up in Perplexity there, with brands and citations listed. Those ranks are a snapshot from this session, not a later result. For citation records turned into a review plan in Codex, see CrowdReply MCP with Codex: Citation Data to Review Plans. For keyword data over MCP in Claude Code, see DataForSEO MCP: build an AI SEO agent in Claude Code.
4. Install the Rank Prompt MCP in Claude Code
To edit pages from that report, he opens Settings in the bottom left, then MCP. Rank Prompt shows install instructions for Claude Code and Codex. He pastes the install prompt into Claude. The recording does not speak that prompt in full, so this page does not reconstruct it. Claude Code then adds the Rank Prompt HTTP MCP server. He says it is in Claude.json and will load in all projects.
To confirm it loaded, he runs:
/mcpHe scrolls for the Rank Prompt MCP, exits, reopens the session, and sees it listed. First use of Claude Code in this flow still needs an API key, which is the next section.
5. Authenticate with an API key
The first data pull fails. Rank Prompt asks him to sign in. He tells Claude Code:
auth rank promptA browser auth is not enough in this take. Rank Prompt then asks for an API key. He creates one named Claude, with read all access and write access, scoped to VisiScan. He copies the secret key into Claude Code, grants access, and authentication succeeds. Treat that key like a password. Do not paste it into a shared chat.
6. Audit and fix website content
With the MCP connected he asks Claude Code to pull prompts VisiScan is not ranking for and to start implementing fixes. Spoken first request:
Can you go ahead and pull in all of the prompts for using rank prompt that I'm not currently ranking for and start implementing fixes?He already has about 50 blog articles on the site. After auth, a shorter follow-up is:
go ahead and pull all of the ranked promptsThen:
show me what to fixClaude Code reports Rank Prompt has 14 of its 15 rank prompts, checks which pages in the repo it could fix, and notes the site already has pages for those topics. The returned list includes missing compare pages and an onsite answer-first rewrite. It asks which to do first, and says the pick is either two or three on a branch, with one as a free add-on. He tells it to do two and three.
Inside Rank Prompt itself he can click Improve article for a step-by-step rewrite, or keep using the MCP so Claude Code edits the repo he can push to GitHub. The dashboard also shows priority, a detailed audit, quick wins, content gaps and source insight. Review those edits before you deploy. Another MCP-to-agent loop, with a report-first step, is in the Trakkr and Codex visibility workflow.
7. Backlink outreach and email finder
The sources view in this recording lists 432 cited pages. Zapier is the number one result. Find contacts on that row uses one credit from the monthly 1,000. Rank Prompt returns an email, can generate a backlink-request message, and can send it after a mailbox is connected.
He connects Gmail, adds a display name, sets a daily send limit and a signature, and marks that account as default. The generated subject line in this take is "a useful addition to your AI visibility guide", with a link to his site in the body. He repeats find-contact and generate-email on another cited site. Later, contacts are filled for 30 of those sites, and Gmail shows the formatted outreach going out.
This page does not paste description-footer affiliate URLs. Apollo is named on camera as a more expensive alternative he is not using here. For other outreach-from-an-agent setups, see CrowdReply MCP with Codex.
8. Deploy the recorded fixes
While outreach runs, Claude Code finishes items two and three, including blog edits and the sitemap, and offers to push the branch to GitHub. He says push it. On camera that push is meant to deploy to the live VisiScan site. The recording says it added five new comparison pages. Claude Code then warns that this will not change Rank Prompt's numbers by itself, and that on-site changes help most once off-site mentions and Perplexity recrawls start to land.
Follow-up spoken in the recording:
What other fixes do you recommend based on all of my ranked prompts and how can I improve my overall AI search visibility across the site?The later report says only one of the 15 tracked prompts names the site: Perplexity at position three on the free-scan prompt, with overall visibility at one. He then says:
Just go ahead and implement all of the fixes, please.For remaining contacts he asks:
Go ahead and create the drafts for the outreach for me using the MCP.Claude Code notes that four outreach emails already went out from the mailbox and starts drafting the rest. A send request follows:
Can you automatically send off these drafts to the appropriate contacts using the MCP?He opens /mcp again. The recording says the server has about 94 tools. Claude Code still cannot send from the terminal in this take. It recommends opening the email hub in the Rank Prompt app and pasting each draft. He sends from the site instead. Nothing in the video measures a later visibility lift. Treat the push as deployed code, not as a proven AI-search win.
Watch the walkthrough
FAQ
Did Rank Prompt sponsor this video?
Yes. Rank Prompt sponsored this video. This page uses the public product site with no tracking parameters.
Did the recording show the MCP install prompt in full?
No. Mike opened Settings, then MCP, and pasted an install prompt into Claude Code. The recording did not speak that prompt's full text, so this page does not reconstruct it.
Could Claude Code send the Rank Prompt outreach emails?
Not in this recording. After checking /mcp, Claude Code said the fastest way to send three drafts was to open the email hub in the Rank Prompt app and paste each draft. Four emails had already gone out from the mailbox earlier in the session.
Did the on-site edits change Rank Prompt numbers on the spot?
No. Claude Code said the on-site changes would not change Rank Prompt's numbers by themselves, and that they help most once off-site mentions and Perplexity recrawls start to land.
Sources
- Video: Your Business Is Invisible to AI Search (Here's the Fix) (12 November 2026)
- Rank Prompt: https://www.rankprompt.com
You may also like
Build the next one with AI Income Lab
Step-by-step courses, reusable templates, and a community that answers when you get stuck.
Members get the Claude Code & Skills and Codex Tutorials & Courses lessons.
- Courses for builds like this one
- Copy-ready templates
- Ask questions while you build
- Weekly coaching on VIP
