Ranksmile runs an MCP server, so an assistant you already use can read what is in your account and reason about it with you.
https://app.ranksmile.pl/mcpPer-client setup, including Claude Desktop, Claude Code and Cursor, is on the MCP page.
Claude Code plugin
In Claude Code, you can install the Ranksmile plugin instead of adding the URL by hand. It connects the same MCP server and adds three workflows that tell the assistant which tools to use and in what order:
- diagnose-low-score: why an article scores low, and the few edits that would raise it most.
- content-refresh: pages close to Google's first page (positions 8 to 20 in Search Console), with a refresh plan for each.
- visibility-gaps: questions where AI answers name competitors instead of you, and what to publish to close the gap.
/plugin marketplace add Idzikowski20/ranksmile-plugin
/plugin install ranksmile@ranksmileThe same command is in the app under Settings > Integrations > MCP. With Read and write on the consent screen (the default when the plugin asks for it), the workflows save what they found to your research log, so the next run doesn't pay for the same research again.
How access works
The client registers itself, you approve it once in the browser, and the token it receives is scoped to your account. There is no API key to paste anywhere and nothing to rotate by hand.
An assistant sees at most the Brand Spaces the approving account can open, and only the ones you choose when you approve it. Connecting a client does not widen anyone's access, and it does not create a shared credential.
What it can read
Your workspaces, your articles and their scores, the breakdown behind a score, the Auto-Optimize history for a piece and the recent generation jobs behind it, plus AI visibility, tracked keyword positions, Search Console performance, the site audit, and your brand notes and research log.
On the consent screen you choose which workspaces and tool categories the assistant can use: all of them, or only the ones you pick. Read and write also lets it add to your brand notes and the research log; Read only keeps it to reading. It can't edit, publish or delete articles either way.
Writing articles stays in the app, where a person approves it. An assistant reading your scores is useful; an assistant rewriting published pages because it misread an instruction is not a trade worth making.
What it is actually good for
Debugging, mostly. The honest use case is not "ask the assistant to do SEO", it is "ask the assistant why this specific thing looks wrong".
A page scores badly and you cannot see why. The assistant reads the full breakdown rather than guessing from the text, and can explain which concepts are missing and how far off the structure is.
Auto-Optimize appears to have changed nothing. The history holds the before and after and the outcome. The assistant reads it and tells you which of the ordinary reasons applies. See Auto-Optimize.
Something generated came out wrong. The job record shows what the generator was working from and what it returned, which is usually the whole explanation.
"Why does article 412 score lower than 388" gets a useful answer. "Improve my SEO" gets a summary of things you already know. The tools return specifics, so the questions should be specific.
What it is not
It is not an API. There is no REST interface, no API key, and no way to drive the product programmatically. MCP is a read path for assistants, not an integration surface for building on.
Where it connects
Content score and Auto-Optimize produce most of what the tools return. Smily AI is the assistant inside the app, which is a different thing and can do more.
Need help?
Write to us and a person answers. See the frequently asked questions first if your question is a common one, or get in touch for anything else.