How do I set up the GetMint MCP?

The GetMint MCP lets your AI agent read your GetMint data and run Content Studio directly. Once connected, you can ask Claude (or any MCP-capable client) questions like "how did my visibility on topic X change this week?" and the agent pulls the answer from your live project — or kick off a brief and content generation without leaving the chat.

The endpoint

There's one URL for everything:

https://mcp.getmint.ai/mcp

It uses the MCP Streamable HTTP transport (spec 2025-03-26).

Pick an auth method

Two ways to authenticate, both granting access to the same surfaces:

Both coexist; adopting OAuth doesn't break existing keys.

Heads up: only an organization Owner or Admin can authorize a connector — its token can read and write across all of your organization's data. Editors and Viewers will see a "you can't authorize this connection" screen; ask an owner/admin to set it up.

Connect from Claude.ai (OAuth — recommended)

1. In Claude.ai, open Settings → Connectors → Add custom connector.
2. Paste the URL https://mcp.getmint.ai/mcp and click Connect. There's no key field — Claude connects over OAuth.
3. Sign in to GetMint and approve the consent screen. The connector requests read access to Insights and read/write access to Content Studio briefs and documents.
4. Open any new chat — the GetMint connector now shows under the Tools menu. Toggle it on and ask a question that needs your project data.

Connect from ChatGPT

ChatGPT's custom connector UI accepts both auth methods — pick whichever fits.

1. Settings → Connectors → Add custom connector, paste https://mcp.getmint.ai/mcp.
2. Choose your auth method:
   • OAuth — sign in to GetMint and approve the consent screen, same as the Claude.ai flow.
   • API key — set Authorization: Bearer mint_live_YOUR_KEY_HERE.

Connect from Claude Desktop

Add the server to ~/Library/Application Support/Claude/claude_desktop_config.json:

{
  "mcpServers": {
    "getmint": {
      "type": "http",
      "url": "https://mcp.getmint.ai/mcp",
      "headers": { "Authorization": "Bearer mint_live_YOUR_KEY_HERE" }
    }
  }
}

Restart Claude Desktop. The getmint server appears alongside any other MCP servers you have configured.

Connect from Claude Code

Add the server to your project's .mcp.json (or ~/.claude.json under mcpServers for a user-level entry):

{
  "mcpServers": {
    "getmint": {
      "type": "http",
      "url": "https://mcp.getmint.ai/mcp",
      "headers": { "Authorization": "Bearer mint_live_YOUR_KEY_HERE" }
    }
  }
}

The "type": "http" field is required — without it Claude Code treats the entry as a stdio server and ignores url/headers.

Restart Claude Code and run /mcp to confirm the server is connected.

Connect from Cursor

Settings → MCP → Add Server:

{
  "getmint": {
    "url": "https://mcp.getmint.ai/mcp",
    "headers": { "Authorization": "Bearer mint_live_YOUR_KEY_HERE" }
  }
}

Where to get an API key

In GetMint, open Settings → API Keys. Keys start with mint_live_ . One key gives you access to the REST API and the full MCP toolset.

Treat keys like production secrets. They can trigger credit-consuming actions (brief creation, content generation). Use one key per integration, never commit them, and rotate immediately on exposure from the same screen.

What you can do with it

The single /mcp endpoint exposes two tool groups:

• Insights — read-only. List and inspect domains and topics; pull Visibility Score, Share of Voice, Sentiment, Brand Alignment, competition, and the top brands cited per topic.
• Content Studio — read/write, consumes generation credits. List and create briefs, check brief progress, fetch documents, trigger content generation, and retrieve the finished article.

What's intentionally out of scope (for now): publishing content from MCP, plan/billing changes.

Verify it works

In a new chat with the connector enabled, ask:

"List the topics tracked on my main domain."

You should see the assistant call list_topics (Claude.ai shows tool calls inline). The result should match what you see in your GetMint dashboard. If the assistant says it can't see your data, re-check that the connector is toggled on for the chat and that your authorization hasn't expired.

Rate limits

All MCP calls count against the same global limit as the REST API: 200 requests per minute per key. A single MCP tool call is typically one API request, so MCP does not multiply your request budget.

Troubleshooting

• The connector doesn't appear in Tools. Reload the chat tab; Claude.ai picks up new connectors on the next chat session.
• 401 Unauthorized mid-session. Your key was revoked or your OAuth token expired. Replace the key in your client config (header-based clients) or reconnect from Settings → API Access → Connected Apps (OAuth).
• "You can't authorize this connection." OAuth connectors require Owner or Admin role. Ask an owner/admin to set it up — once authorized, every member with the connector enabled in their client can use it.
• The assistant returns stale data. GetMint refreshes on its own cadence (weekly on Starter). If you just changed something in the UI and don't see it via MCP, give it a few minutes.

Key takeaways

• One endpoint: https://mcp.getmint.ai/mcp — exposes Insights and Content Studio together.
• Two auth methods that coexist: API key (header-based clients) or OAuth (Claude.ai, ChatGPT). Adopting one doesn't break the other.
• OAuth needs an Owner/Admin to authorize the first time; API keys can be used by anyone you share them with.
• Same rate limit as REST: 200 req/min/key. Same key, two surfaces.
• Risky writes stay in the UI — MCP gives you reads and Content Studio writes, not destructive operations.