AI Connector (MCP)
Miraqo, right inside your AI assistant chat.
Connect Claude, ChatGPT, Gemini, Claude Code or Cursor to your SEO data and query it in natural language, without opening the dashboard.
The AI connector brings Miraqo's data into the assistant you already use. MCP, Model Context Protocol, is the standard an application uses to expose its tools to a model: once connected, the assistant answers questions like «which keywords did I lose this week?» or «summarise the critical audit issues» by reading the project data: rankings, audit and Core Web Vitals, backlinks, AI Visibility and GEO Audit, competitors, Search Console and Topical Cluster. claude.ai, ChatGPT and Gemini connect with the Miraqo account, no token needed; Claude Code and Cursor with a personal token.
Reads do not use the plan quota. Paid actions, such as running an audit or a keyword research, start from the chat, use the quota as they would in the app and have a dedicated monthly cap. The connector follows the same permissions as the app, connected apps and tokens are revoked one by one, and the same tools are available as a documented REST API.
AI Connector (MCP): Miraqo's SEO data inside the AI assistant you already use
Questions, in chat
The assistant reads the project data and answers in natural language.
- All the project's dataKeywords and ranking history, YouTube rankings, audit issues, pages, redirects and Core Web Vitals, backlinks, AI Visibility and GEO Audit, competitors and keyword gap, Search Console, Topical Cluster.
- Answers from real dataThe numbers come from the project, not from the assistant's guess.
- Several projects and organisationsThe list of accessible projects, so anyone working for several clients says which one they mean.
- Reads without quotaQuerying the data does not use the plan quota, however often it is needed.
Actions from the chat
Some operations start straight from the conversation.
- Seven actionsRun an audit, a keyword research, a domain analysis or an AI Visibility check, add an AI prompt, refine the clusters, generate a semantic bridge.
- Same usage as the appPaid actions use the quota just as when they start from the app.
- Dedicated monthly capActions through the connector have their own monthly limit, separate from reads.
- Visible usageThe assistant can read the remaining credits before starting an operation.
How it connects
One address, two ways to authorise access.
- A single addresshttps://app.miraqo.io/mcp, the same for every assistant.
- claude.ai, ChatGPT and GeminiThe address is added as a connector and authorised with the Miraqo account, no token needed.
- Claude Code and CursorA personal token, with the command or the mcp.json snippet ready to paste.
- REST APIThe same tools as a documented API, for your own integrations.
Permissions and security
The connector sees what the user sees in the app, nothing more.
- Same permissions as the appEach user reaches only the projects and sections they are allowed: a client with limited access stays limited in chat too.
- Revocable connected appsEvery authorised assistant appears under «Connected apps» and is disconnected with one click.
- Tokens under controlA token is shown only once, up to five can be active and each one is revoked on its own.
- Third-party data as dataThe connector warns the assistant that titles, AI citations and anchor text come from external sites and must not be followed as instructions.
AI Connector (MCP): how it works
- 01
One address for every assistant
claude.ai, ChatGPT and Gemini are authorised with the Miraqo account; Claude Code and Cursor use a personal token.
- 02
The assistant reads the project data
Keywords, audit, backlinks, AI Visibility, competitors, Search Console and Topical Cluster become answers in natural language.
- 03
Actions when needed
Audits, keyword research and checks start from the chat, with visible usage and a dedicated monthly cap.
Use cases
AI Connector (MCP): when to use it
- Ask the AI for a project summary without opening the dashboard
- Trigger an audit or keyword research while working inside Claude Code or Cursor
- Build custom automations with the same connector's REST API
Questions about ai connector (mcp)
No. On claude.ai, ChatGPT and Gemini the connector address is added and access is authorised with the Miraqo account, without generating a token. Claude Code and Cursor use a personal token with a ready-made snippet.
Generate a personal token from Account Settings and paste it into the configuration snippet shown on the same screen.
No. Reads (rankings, audit, backlinks, AI Visibility, competitors, Search Console) do not use the plan quota. Paid actions, such as running an audit, use it as they would in the app and also have a dedicated monthly cap for connector actions.
The connector applies the same permissions as the app: each user sees only the projects and sections they have access to. Authorised apps are revoked from «Connected apps», tokens (up to five active) are shown only once and revoked one by one, and the assistant is warned that content from third-party sites is data, not instructions.
Yes, the same tools behind the connector are also available as a documented REST API, for custom integrations.
The AI Connector (MCP) guide in the Help Center covers step-by-step setup, including ready-to-paste snippets for Claude Code and Cursor.
Related tools
AI Visibility Monitor whether ChatGPT, Gemini, Perplexity, Google AI Overviews and AI Mode cite your brand — and find out how they describe it. Keyword Research Related keywords, keywords by domain and competitor gaps, with volume, CPC and competition from Google data. Site Audit Miraqo crawls your site and gives you a health score with actions sorted by impact. See all featuresAI Connector (MCP): step-by-step guides
Start monitoring
your projects today.
Create a project, add keywords, see real positions. Free to start.
Start free