Run ScoutRival from your AI assistant.
The MCP connection lets Claude, ChatGPT, Cursor and other AI clients read your ScoutRival data and run its tools for you — today's brief, competitor moves, AI-visibility, content, one-click drafts — without leaving the chat.
MCP — the Model Context Protocol — is an open standard that lets an AI assistant use tools from another product. When you connect ScoutRival, your assistant can pull your daily brief, list what your competitors did this week, read your AI-visibility score, write a post in your brand voice, and save it to WordPress as a draft — all from the same chat you already work in.
You connect one server — https://user.scoutrival.com/api/mcp — and your assistant does the rest. There is nothing to install and no separate app to open. Read tools cost nothing; a few tools that generate content or run a check spend credits, exactly as they do inside ScoutRival.
What this is, in one paragraph
Your assistant already answers questions and writes text. The MCP connection gives it a set of ScoutRival tools on top of that — a way to fetch your real data and act on your account. You stay in control: you approve each action in your assistant, and you decide how much access a connection has. Nothing runs on a schedule; the assistant only does what you ask, when you ask.
What you can do once it is connected
- Read your morning brief. "Give me today's brief for my brand" returns the verdict, what each rival did, and the suggested posts — the same brief you see in the app.
- Check where AI names you. "How visible am I in AI answers, and what's the highest-leverage fix?" pulls your latest AI-visibility check and its recommended actions.
- Turn a gap into content. "My competitors rank for emergency roof repair and I don't — write the article and save it as a WordPress draft."
- Watch your rivals. "Summarise what my five competitors shipped this week."
- Run a check on demand. "Run a site AI-visibility audit at the essential depth and tell me the three things blocking citations."
The read tools return the full picture — your brief, competitors, AI-visibility, SEO, traffic and content — not a summary. See the tool reference for the complete list.
Which assistants work
MCP is a feature of the assistant, not the AI model. You connect through a client that supports it, and the model runs inside that client. Two ways to connect, depending on the client:
- Paste a token — Claude Desktop, Claude Code and Cursor take a connection token in their settings.
- One-click sign-in (OAuth) — Claude.ai on the web and ChatGPT connect by signing in to ScoutRival, no token to copy.
Grok and the Gemini CLI also connect a custom server. Models without their own connector — such as DeepSeek, Kimi, or the Gemini app — can still be used: run them as the model inside a client that supports MCP. Each client has its own short guide in this section.
Plan and credits
- MCP is available on Starter and up. Free plans don't include it — you'll see an upgrade prompt on the token page.
- Reading your data is free. Listing brands, reading briefs, checking AI-visibility or SEO — none of these spend credits.
- Generating and running cost credits — the same price as in the app (a social post is 2 credits, a full article 10 to 50, an AI-visibility check 2 per answer). The tool tells the assistant what it charged, and you can ask for a price before you run.
- Publishing to WordPress via MCP always creates a draft, never a live post — you publish it yourself.
Where to start
The quick start gets you connected in about five minutes with a read-only token. From there, pick your assistant's guide and, when you want to generate content, read scopes and access levels to grant a little more.