Can you check your AI visibility from inside Claude or ChatGPT?

Yes. Vidrys runs a Model Context Protocol (MCP) server, so Claude, ChatGPT, Cursor or Codex can read your AI visibility data and act on it in conversation: 37 tools, 32 that read and 5 that write. It shows which buyer questions ChatGPT, Perplexity and Google AI Overviews answer with your brand, and who wins the rest. It is included from the Track plan at ₹6,999 a month.

What is an MCP server, and why does it matter for AI visibility?

The Model Context Protocol is an open standard, introduced by Anthropic in November 2024, for giving an AI assistant a defined set of tools it can call. Instead of exporting a report and pasting it into a chat, you connect the assistant to Vidrys once, and it fetches the live numbers itself whenever you ask.

That matters because AI visibility work is a loop of small questions: did we move this week, which answer did we lose, what did it cite, what should we write. An assistant that can call the data answers each one in seconds, in the tool you already work in.

What can you ask it?

Each of these resolves to one or two tool calls:

• “How did our AI visibility move this week, and which engine got worse?”
• “Which buyer questions are we losing on ChatGPT, and who wins them instead?”
• “Why did we lose that Perplexity answer? Give me the reasons and the actions.”
• “Which third-party sites do the answers we lose keep citing?”
• “What should we publish next, and what does the brief say?”

What can the write tools change?

The 5 write tools add or edit tracked prompts, move content opportunities and queue tasks between stages, and generate a draft. Only generating a draft costs anything: one content credit, and the tool asks for explicit confirmation first, so an assistant cannot spend credits without a person approving. A read-only key cannot call any write tool.

How do you connect it?

1. 1Create an API key on the Developers page of the Vidrys dashboard, or use a sign-in connection where your client supports one.
2. 2Add the server URL https://api.vidrys.com/mcp to your assistant's connector or MCP settings.
3. 3Ask your first question. A session starts with list_projects, which supplies the project id every other tool needs.

The MCP quickstart has the exact configuration for Claude Code, Codex, Cursor, claude.ai and ChatGPT, and authentication covers keys, sign-in connections and how to revoke either.

Which Vidrys plans include the MCP server?

Prices are in Indian rupees, exclude GST, and were checked on 27 October 2026. The workspace also has to be approved before the server accepts its key. Reads cost no credits but share the plan's per-minute API budget: 60 calls a minute on Track and Grow, 300 on Scale and 600 on Custom.

Custom is quoted. Full details are on Pricing, and every tool is listed with its fields in the MCP docs.

Is it safe to let an assistant read this data?

AI answers and the pages they cite are text written by other people, so the server returns them as data and never as instructions; the security notes explain how. Each key is scoped to your workspace, and you can revoke a key or a connected app at any time. Read the MCP specification for how the protocol itself works.